Comments by "Lawrence D’Oliveiro" (@lawrencedoliveiro9104) on "Computerphile" channel.

  1. 1:52 “Make sure everyone’s on the same page.” Flips to a different page
    1000
  2. The separation between the S and T servers is the separation between authentication (proving you are who you say you are, done by S) and authorization (what services you are allowed to access, controlled by T). Each one can be updated independently.
    253
  3. 3:37 Note that each λ represents a function definition: so “λx.λy.x + y” is not (directly) a function of two arguments: it is a function of one argument (x) returning a function of another single argument (y) which returns the sum!
    139
  4. Here’s a key acronym to remember about GPUs: “SIMD”. That‘s “Single-Instruction, Multiple-Data”. It has to do with the fact that a GPU can operate on a hundred or a thousand vertices or pixels at once in parallel, but it has to perform exactly the same calculation on all of them. Whereas a single CPU core can be described as “SISD” -- “Single-Instruction, Single-Data”. With multiple CPU cores, you get “MIMD” -- “Multiple-Instruction, Multiple-Data”, where each instruction sequence can be doing entirely different things to different data. Or in other words, multithreading. So even with all their massive parallelism, GPUs are still effectively single-threaded.
    130
  5. 2:56 The key lies in observing that each half of the original data block does not go through the F-block to get to the next stage. The only use of the output of the F-blocks is to xor with data blocks. And xor is reversible: xoring with the same bitstream twice gives you back the original bitstream. That’s why the same sequence of bit-mashing works for both encryption and decryption. It really is that simple.
    119
  6. The irony being that he was a linguist, but linguists found little or no use for his mathematical innovation. The computer scientists, on the other hand, lapped it up.
    101
  7. Unless the question was “why is half your brain missing?”
    68
  8. 7:50 Some people might be accustomed to other VCSes which automatically assume that every change you have made to your working copy is to be included in the next commit (e.g. Subversion works this way). Git doesn’t do this -- it requires you to explicitly add changes to the “index” to be included in the commit. This is because Git recognizes that you often make multiple sets of changes -- for example, while fixing one thing, you notice something else that needs fixing, that is quite unrelated. So when you actually get round to committing your changes, you can pick and choose which ones go into the same commit, and which ones to leave for another separate commit. And some things, like debug message lines, can be left out altogether.
    62
  9. Some people, faced with the problem of comprehending a presentation on regular expressions, like to distract themselves with a well-known and overused Jamie Zawinski quote. Now they have two problems ...
    56
  10. I remember in the mid-1990s, working on the first automation system for the NZ DNS registry, and learning Perl at the same time. The script we started with (from the Australian registry) was storing 2-digit year numbers. A (presumably more experienced) colleague immediately said not to worry about it, that there was no way our particular software would still be in use by the end of the century. I disagreed and insisted that years be stored as 4 digits. That was purely out of instinct, which turned out to be correct.
    56
  11. 1:33 Two keys, not three for triple DES. That’s what it says in Tanenbaum’s Computer Networks text, 4th ed, page 740. You encrypt with key K₁, decrypt(!) with K₂, then encrypt again with K₁.
    55
  12. 1:16 Space before colon, but not after comma. ‌twitch‌
    49
  13. My attitude to comments is that they shouldn’t say just what the code does (the code itself does that), but why the code is doing what it’s doing. Perhaps a comment on the regular expression could be expressed on the same basis.
    46
  14. That, and the fact that their “end-to-end” encryption wasn’t end-to-end at all, it was end-to-middle, middle-to-end. They were also criticized for using AES-128 instead of AES-256, but really AES-128 is fine.
    42
  15. 10:02 And one nice thing about scan codes is they don’t distinguish between modifier and non-modifier keys; it’s just “key «n» has been pressed” and “key «n» has been released”, it’s up to the software to decide if key «n» is a modifier key or not. And you can distinguish between e.g. left-shift and right-shift (they have different scan codes), and implement other fun custom behaviours.
    40
  16. 9:54 Back when Ethernet was expensive, Apple developed its own, cheaper† “LocalTalk” layer-2 connection system for its AppleTalk protocol stack. Besides the much lower bandwidth and lower-spec wiring, another way they kept the cost down was to omit the collision-detection hardware. An Ethernet node could notice if someone else started transmitting while it was doing so, but LocalTalk could not. Instead of being “CSMA/CD” (the “CD” standing for “Collection Detection”), LocalTalk was “CSMA/CA” (the “CA” standing for “Collision Avoidance”). †Yes, “Apple” and “cheaper” in the same sentence. Not something that happened very often, even back then.
    38
  17. 3:56 ALGOL 68 had its own terminology for so many things: types were called “modes”, addresses/pointers/references were called “names”, and instead of saying “input/output” or “I/O”, they coined the term “transput”.
    35
  18. 6:10 A common use for mouse-moved events is changing the mouse pointer, e.g. making it an I-beam pointer when the user moves into a text field, to indicate that they can type there.
    29
  19. CP/M was strongly influenced by DEC minicomputer timesharing operating systems of the 1970s. The name “PIP” (“Peripheral Interchange Program”) for the file management utility is exactly that of a corresponding DEC utility found across several of its OSes. Only of course the DEC versions had a lot more functionality. One thing, though: DEC OSes had multi-character device names; pity CP/M had to “simplify” them down to a single letter.
    29
  20. 5:47 If you think about it, the OSI seven-layer model included a “presentation” layer, between “transport” and “application”, that nobody could fit into the reality of TCP/IP very well. But SSL/TLS fills that layer very nicely.
    28
  21. At least XML allows comments.
    27
  22. 9:41 Remember, none of these public keys is confidential. The problem is, you have to trust they are authentic, that they really came from the people they say they’re from.
    27
  23. A couple of significant notational differences between C and Pascal are, when declaring variables, C inherits the Algol syntax of type-spec before variable name: “«type-spec» «var-name»” whereas Pascal invented its own syntax: “«var-name» : «type-spec»”. And also pointer dereferencing: C uses a prefix operator “*«pointer-val»” whereas Pascal has a postfix one: “«pointer-val»^”. This may seem very tomayto-tomahto, but consider how C++, which inherits C syntax lock stock and barrel, is now trying to deal with newer features involving dependencies between types, for example between the arguments to a function and its result type. It now finds that, in addition to the C syntax of putting the return type before the function name, it also has to add an alternative syntax where the return type comes after the argument list, à la Pascal. As for the pointer issue, this may not occur very often, but back in my Mac programming days, more than once I had to write an expression like “thePort^^.theGDevice^^.gdPMap^^.pmTable”, which in C would have had to become “(**(**(**thePort).theGDevice).gdPMap).pmTable”. (These were structures defined for interfacing to the Mac system, not created by me.) Which would you prefer?
    27
  24. 20:23 This is where “quality of service” (“QoS”) comes into play. TCP/IP allows for marking packets with indications as to whether timeliness of delivery is more important than reliability of delivery: that is, if a packet is taking too long to be delivered, it might as well be dropped rather than delivered too late to be of use.
    27
  25. You didn’t mention Ada, which was an offshoot of Pascal created as the result of a competition held by the US Department of Defense, to create a language suitable for doing systems programming for safety-critical and mission-critical systems. There is still Ada code in use today, for example controlling the life-support systems on the International Space Station (running on Intel 386 processors, with no Microsoft code anywhere in sight).
    25
  26. In his honour, instead of #hashtags in future, let us have £poundtags.
    24
  27.  @IceMetalPunk  Code can only do what it does. Comments that describe what the code does are obviously superfluous, but ones that explain why it is doing what it is doing should be essential. How could code ever be “self-explanatory” in that way? It cannot.
    23
  28. The three basic objects in Git are blobs, trees and commits. A blob represents the contents of a file, and is identified by the SHA-256 hash of its contents. A tree associates names with blobs, and represents a snapshot of source tree contents at a point in time. A commit holds a pointer to a tree, and to zero or more parent commits, as well as information about who did the commit and when, an explanatory comment etc. Trees and commits are also represented by SHA-256 hashes of their contents. This is why a commit holds pointers to its parents, not to its children; because once the commit has been created, it cannot be changed (e.g. to add child pointers) without losing its original SHA-256 ID. Branches and tags are just pointers to particular commits. They have no associated SHA-256 hashes, so they can be added, removed and changed at any time.
    23
  29. 7:04 Notice the only kind of recursion allowed is tail recursion—there is nothing after the B that occurs in the substitution for B itself. This can be implemented with simple iteration, without a stack. Otherwise, it would no longer be a Type 3 grammar.
    22
  30. 3:41 Tip: You can do “git show --raw «commit»” to get a list of the files affected by that commit. You can similarly list the files affected by each entry in the log with “git log --raw”.
    21
  31. 6:57 There is another technique, called “pipelining”. Or, in industrial terms, you set up an “assembly line”, where one person just butters the bread, and passes the slices onto the next person to put in the cheese and slap them together. Modern CPUs use this technique as well as multiple processors. And then there’s also caching.
    20
  32. If there was any “spite”, it was not on his part, but on that of Larry McVoy in cancelling their BitKeeper licence.
    19
  33. 27:05 That is the inevitable fate of every GUI architecture, to get more and more complicated over time. There are no “simple, efficient” GUIs.
    18
  34. 5:19 Isn’t this also where the “CIRC” (“Cross-Interleave Redundancy Checking”) comes in? The bits are arranged in a pseudorandom scrambled pattern (which is unscrambled as part of the decoding process), so that any contiguous run of errors on the disc is spread across a larger area after unscrambling and affects a smaller proportion of bits within that area.
    17
  35. Better for number-crunching. Wipes crumbs from mouth
    17
  36. The term for those artifacts is “mosquitoing”. They are the image equivalent of “ringing” in digital audio.
    15
  37. 2:10 The later term for this separate processor on the ARPAnet was “Interface Message Processor” or “IMP”. It was probably necessary because IBM mainframes did not like to be interrupted for every little bit or byte of data -- their I/O was done through elaborate “channel” processors which could be given sequences of operations to perform without bothering the CPU until the sequence was complete -- or some unusual situation arose (e.g. an I/O error). This probably also eased the problem of implementing the network, because the IBM OS wouldn’t have been interactive. IBM preferred terminals that operated in “block mode”, where you had fields defined on-screen by previous computer output, that you filled in locally, and nothing got sent back to the mainframe until you hit the “Send” key.
    15
  38.  @gordonrichardson2972  The 486SX was a completely cynical marketing ploy. It lacked the (functioning) floating-point hardware, but you could buy a 487SX chip to add this. However, the 487SX was more than just a floating-point coprocessor: it was actually a complete functioning CPU. Plugging it in caused the 486SX to turn itself off and hand over all functions to the 487SX!
    14
  39. 6:14 Unfortunately, the Java specification is barely smaller than the C++ one. So much for trying to “simplify” -- it just introduced its own complications ...
    14
  40. Newton’s Third Law, action = reaction; contemplate the Zen of physics: when you a press a key, the key presses back.
    14
  41. Thumbs down if your calendar still goes by “AD”.
    13
  42. 4:05 Not just a security issue: think about languages that require a front-end processor (e.g. Chinese/Japanese/Korean) to translate keystrokes into a large character set: putting this in userland makes it more flexible, replaceable, configurable etc.
    13
  43. 8:34 So lex was written by a guy named Lesk. So near, and yet so far ...
    12
  44. I’ve wondered why telephone numbers don’t have check digits. It would reduce the incidence of wrong numbers by 90%. Yes, it would make them longer. That mattered more back when they might only have been 5-6 digits. Now that they are commonly 7-8 digits or longer (particularly for mobile phone users), it seems less of an issue. Come to think of it, why are we still using telephone numbers? They’re an idea that dates from the 19th century. Why don’t we use 20th century Internet-based technologies, such as the Domain Name System? Use names, instead of numbers, to connect to people.
    12
  45. 6:19 I remember a description from the time that “pointers are the gotos of data structures”. And guess what: Pascal had gotos as well. Not even restricted gotos: they were a full implementation, including being able to jump out of an inner procedure/function call into an outer one. This was, of course, a way of doing exceptions before languages had the concept of “exceptions”.
    12
  46. 5:22 You can set your working directory to any subdirectory of a project directory. Git will automatically go up into parent directories until it finds one containing a .git subdirectory, or until it hits a filesystem boundary, at which point it gives up.
    12
  47. The Apollo Guidance Computer used “rope” memory for its ROM -- the bits were encoded in the weave.
    12
  48.  @MatthewStinar  COBOL didn’t really have the concept of arithmetic in bits. It preferred to do it in decimal, with a precision of 18 decimal digits.
    12
  49. The paradox of modern cache-intensive CPU architectures: RAM is cheap, but accessing RAM is expensive.
    12
  50. 7:17 If you operate in high-dynamic range formats like EXR, which has floating-point pixel components, you have room to both overshoot and undershoot. Then you can renormalize the image using a filmic response curve or something without clipping.
    12
  51. The context for that comment was I think the phenomenon that people who have just discovered regexes tend to (over)use them for everything. One horrendous example was a regex for parsing e-mail addresses.
    11
  52. Version control has evolved a lot over the decades. In the beginning, it was thought that it was vital to lock files when somebody checked them out, because if two people made simultaneous changes to the same file, the world would end or something.
    11
  53. Don’t spend it all at once.
    11
  54. Part of it is just text pointers, which are easy enough to examine with regular file tools. E.g. the HEAD file just contains the name of the “current” branch. Unless you have checked out a commit in “headless” mode (what I like to call “Marie Antoinette” mode) ...
    11
  55. 11:17 There is a thing called “QoS” (“Quality of Service”). This has to do with the differing needs of different Internet applications: for example, file transfer requires maximum reliability, even if packets get held up occasionally, while telephony and video streaming require timeliness, and can tolerate the loss of a packet here and there. But this is something that needs to be explicitly specified by the applications at the endpoints, the ISPs have no business second-guessing these requirements. The creators and users of the apps are in the best position to know what they want: those who run the network are not.
    11
  56. 9:53 Some people think that stuttering might be caused by a neural analogue to this sort of phenomenon.
    10
  57. Even Apple don’t want to use it any more.
    10
  58. 1:40 I got hold of a copy of the ALGOL-68 spec when I started first year at Uni, and it just blew my mind. I was quite disappointed later to discover that most computer scientists considered Van Wijngaarden grammars to be just too complicated, and so they stuck to simpler, less formal, not to say cruder, ways of defining their languages.
    10
  59.  @MattExzy  It did happen, albeit (mostly) not in a big way. There were instances of system failures from about early-to-mid 1999 onwards, certainly that I can remember. For example, a system for issuing cards with a 6-month expiry, which couldn’t handle expiry dates past 31st December 1999.
    10
  60. But didn’t you have to install the “run” button? And connect it to ... whatever it’s connected to?
    10
  61. Yes, but the vectors that your typical present-day CPU operates on are short ones, with something like 4 or 8 elements at most.
    10
  62. 3:15 Note for Transatlantic folks: “Heath Robinson” (UK) ⇔ “Rube Goldberg” (US).
    10
  63. 2:57 One subtle point worth noting is that the GUI framework usually handles keyboard autorepeat. That is, when you ask for another event, no “real’ event is pending, and a key is still down from the previous one, and the autorepeat interval has elapsed, then the framework will generate an autorepeat key-down event for you. Note that these autorepeat events are not buffered, so they are only generated as quickly as your code can handle them. Thus, autorepeat can never run away from your program’s ability to respond!
    10
  64.  @akashchoudhary8162  You could get fooled into accepting messages from a fake sender instead. So you need to obtain the public key from a trusted source.
    10
  65. That was discussed in part 1. It’s why the matrix is scanned, not simply fed with static voltages.
    10
  66. My employer was mainly a VAX/VMS shop back in the day. I remember when the PHONE program made an appearance, and was a novelty for all of ten minutes. Then everybody made sure to add SET BROADCAST/NOPHONE in their login scripts ...
    9
  67. I think it is quite common to clean up and simplify your private commit history before sending a bunch of patches to the central authorities for a project. For example, you probably don’t want them seeing every little typo correction or whitespace fixup as a separate commit.
    9
  68.  @MatthewBishop64  Except that nobody cares about Dotnet Core. If C# could free itself from all these Windows encumbrances, perhaps it could become a useful language.
    9
  69. 0:28 I see some interesting books on GCHQ and the like. No copy of Spycatcher ? Is that still banned in the UK?
    9
  70. And “Fido” (faithful one) I think was a Latin tradition.
    9
  71. That was after the cancellation of the BitKeeper licence.
    8
  72. 7:53 Don’t forget, server programs are also event-driven, and they often don’t have any kind of user interface at all. Instead, they are responding to events received over client connections.
    8
  73. 9:20 Unlike Subversion, Git does not keep track of what you have checked out at the commit level, only at the branch level. To get around this, it can be helpful to use the “checkout -b” command to create a new, temporary branch pointing to the commit you just checked out. Since this branch is considered “fully merged” (the commit is part of the past history of some branch you really want to keep), you can delete it afterwards with “branch -d” (lowercase d, not uppercase D).
    8
  74. If they’re sharing the same band, then they’re contributing to interference and reducing your bandwidth, even if they’re using an entirely separate protocol.
    8
  75. I think they also introduced “overloading” for multiple definitions of an operator that were chosen according to context. That term is commonplace nowadays.
    8
  76. 8:05 Interestingly, the speed of sound in Earth’s atmosphere at room temperature is almost exactly one millionth of the speed of light.
    8
  77. That video ended with a bit of advice that should have been presented up front: use the existing systems that others have put in place. Even if you want to write your own time-conversion code (and I have), use the Olson database!
    8
  78. Social science is considered a “soft” science. That’s a physicist/mathematician’s way of saying “not a real science”. Remember, this is not just any old “project management” -- this is about the management of software-development projects. If you don’t understand what’s so special about that, you need to read Brooks’ book.
    7
  79. 18:43 The book publishers themselves destroyed a lot of the promise of e-books, by trying to impose restrictions on them above and beyond those on physical paper books. And also trying to charge more for them. I can remember all those years of rising paper book prices in the latter 20th century, which the publishers blamed on the rising cost of paper, ink etc. So you don’t think that moving to a purely electronic format, getting rid of those consumables costs, would see a drop in the price. Not a chance.
    7
  80. 8:56 One of those “couple of other things” in the Amiga was the Copper. This implemented display lists, which allowed it to do fancy animation effects without actually needing to blit large arrays of pixels around. Now that was impressive.
    7
  81. In Windows, the usual answer if some procedure gives you trouble is: “have you tried running it as Administrator?”
    7
  82. As somebody with a Master of Science degree, you can understand why I might disagree.
    7
  83. One of the interesting points about the book is where it discusses improvements in programmer productivity over time. There was an order of magnitude improvement over the decade 1985-1995 (up to the point he did the Anniversary Edition), and he discussed how another order of magnitude might be managed over the subsequent decade. He saw that it was essential to move to ever higher-level languages, which left more of the details to lower layers -- he called this “metaprogramming”. His example of a “metaprogramming” language was AppleScript -- a choice which has not stood the test of time, and one area where the book shows its age. What was more effective at the time was Perl, possibly also Tcl, which have since been joined by others including Ruby, JavaScript/Node, PHP and Python. These are the languages that, in my view, have really made a difference to programming since then.
    7
  84. My first timesharing account as a member of staff was [2, 4]. We didn’t have names in those days; just numbers. Tell that to the kids these days, mumble mumble ...
    7
  85. 17:09 Which brings us to the phenomenon known as “buffer bloat”, where all kinds of devices between your end of the connection and mine have large buffers built into them, just to make some problems easier to solve. But of course this adds latency for real-time applications. So there are people trying to persuade makers of these devices to reduce their buffer sizes. Which requires them to write better code to deal with the problems that they were papering over with large buffers before.
    7
  86. Sorry, I kept saying SHA-256. In fact Git still uses SHA-1. Though there is a plan in progress to move to something more future-proof.
    7
  87. pid content tracker”. He deliberately avoided very elaborate conflict-merging schemes, preferring to concentrate on narrowing down the points of conflict so the human user has just the info they need to resolve the issue, without needless distractions. He said he wouldn’t trust his code to more complex automatic mechanisms that many others were proposing. Yes, “clay” is exactly the word that has come to my mind, too. Having all your work safely stored away in the commit history gives you the confidence to take more risks, knowing that, if you screw up, you can always go back. ;)
    7
  88. Not just editing, but also choreography of Steve’s movements.
    7
  89.  @ChrisJones-rd4wb  The whole concept of “Intellectual Property” actually violates property rights. Think of what it means for the seemingly simple question: “If I buy something, do I own it?”.
    7
  90.  @NoNameAtAll2  It can’t work, because opening/closing tags can be nested inside other opening/closing tags, to arbitrary levels. Regexes cannot handle that.
    7
  91. 7:23 No, the address register is incrementing by 2 each time (note the little gap where the units bit is 0). Machine instructions are multiples of 2 bytes in length, and have to be stored in even addresses.
    6
  92.  @Mtaalas  Luckily, Python’s use of whitespace is pretty much consistent with my formatting conventions anyway. I do still put in “#end” comment lines. If nothing else, it gives Emacs something to jump to, so I can quickly move between the beginning and ending of a block just by looking for lines with the same indentation level.
    6
  93. 5:34 Other OSes had this concept called “overlays”, where you divided up your (large) program into sections such that only some sections were loaded into memory at once. There were complicated mechanisms such that when a procedure in one section made a call into one in another section, the latter would get automatically loaded. But what happened if there wasn’t enough memory? Then some other section had to be thrown out of memory. If no code was currently executing in a suitable section, then fine. But what if there was? Then it would have to be automatically reloaded when control returned to a procedure in there. Or maybe you disallowed unloading sections which held currently-executing procedures. Yes, it all got very complicated. I never wrote an overlaid program in my life. Luckily the 32-bit era arrived just in time.
    6
  94. Rebase is OK for private branches. Once you publish a branch, rebase is going to cause those who pull that branch a lot of grief.
    6
  95. You see now the impetus for RMS to start the GNU project, given the increasingly draconian restrictions on AT&T Unix licensing as time went on. They were bad enough to begin with, but after the breakup of the telephone monopoly, with AT&T now free to fully participate in the computer market, suddenly Unix became a whole lot more commercially valuable to them, and so they became less willing to let Universities and others have it, with full source code, at anything resembling a modest price.
    6
  96. 0:38 “Not that quick” as in “a thousand times slower than secret-key encryption such as with AES”.
    6
  97.  @joshuahillerup4290  The difference between computer science and the branch of maths known as computability theory is that the mathematicians only care whether something is possible or not, while computer scientists actually worry about how efficient it might be to compute. E.g. any encryption algorithm can be cracked in finite time, but if that time is longer than the age of the Universe, then a computer scientist would deem the result to be a useful algorithm, even as the mathematicians would say “meh”.
    6
  98. I wonder why people treat this like some kind of big joke.
    6
  99. I was looking at some statistics a few years ago on contributions to the Linux kernel. At the time there were something like 1000 active contributors, and 10,000 lines of new/changed code flowing into the kernel each day. Which worked out to 10 lines per contributor per day -- which is a figure straight out of Brooks.
    6
  100.  @VAXHeadroom  Have you thought of letting Bitsavers make a scan of it?
    6
  101. On the other hand, why is it that “professional” news networks have such long latency in live connections with reporters in the field, particularly across countries? The anchor says something, the correspondent nods silently for a two or three seconds, then they start speaking. That’s not common with videoconferencing, is it?
    6
  102. 5:37 Assuming the increment of i lies at the bottom of that squiggle somewhere ... otherwise you’ve just written an endless loop.
    6
  103. I think the BSDs are held back by their licensing. This allows proprietary companies to come along and take from them without giving back, and this stunts the growth of the ecosystem. Businesses may not like the sharealike requirement of the GPL, but at least it enforces something closer to a level playing field, and a competitive, thriving marketplace.
    6
  104. 7:29 I had a lecturer at Uni who also had that “NMI” as his middle initial. Of course, in computing, “NMI” more commonly stands for “Non-Maskable Interrupt”.
    6
  105. OK, here’s a fun thought: how would you generalize this idea to a base other than 2? For example, base-26, using the letters of the (uppercase, unaccented) English alphabet? The obvious answer is to do arithmetic modulo 26, assigning numerical values like A = 0 ... Z = 25. The symmetry of xor would no longer apply, so you would need separate encryption and decryption blocks: the encryption blocks would use addition in place of xor, while the decryption ones would subtract.
    5
  106. Was this the early days of cheap knockoff PostScript clones?
    5
  107. 2:55 Nobody seems to understand how to use the Motorola 68k DBcc instructions: they terminate when the count register reaches -1 rather than 0, and everyone thinks this as a misfeature that you have to work around. But it’s a feature! Here’s how you use it: move.w count, d0 bra loop.end loop: ... body of loop ... loop_end: dbra d0, loop You see how that works? You load the exact count into the D-register, not the count - 1 as people often do. And if the count is initially zero, the dbra simply falls through without executing the loop at all! And you get this behaviour without the need for a separate test for the count being equal to 0. DBcc only directly handles a 16-bit count. But it is easy to implement a 32-bit loop count using two DBcc instructions together with a SWAP.
    5
  108.  @timothycurnock9162  How does the key know the difference between a “hack” and a “legitimate decrypt”?
    5
  109. The per-process virtual address space was only 16 bits.
    5
  110. “#include” is such a crude mechanism. Why isn’t Ada-style separation of module interfaces and module bodies more popular? Or look at Python’s module system. The nice thing about both is that they don’t pollute the global namespace without you asking them to.
    5
  111. For me, the moment was seeing “n = n + 1” for the first time.
    5
  112. But in C you are not allowed to put a “;” after “}” to close a block. But you are required to put one after “}” to terminate a struct definition. Not confusing?
    5
  113. Maybe that’s an argument for using names rather than numbers.
    5
  114.  @menachemsalomon  I don’t understand the point of that. It’s just a recipe for needless obfuscation -- and trouble, as you discovered.
    5
  115. 8:11 A blank line at that point would have helped. Just sayin’.
    5
  116. Added complication: those channels overlap. For example, with the original 802.11b on 2.4GHz, there are supposed to be 11 channels, but in fact there is only room for 3 entirely separate channels -- 1, 6 and 11. So all the others overlap one of these (and each other) to quite a large extent.
    5
  117. Von Neumann understood this perfectly well. The only ones who have forgotten tend to use languages like Java or C#. Oh, and PHP.
    5
  118. 4:23 “Kids can understand it” ... while I was doing my MSc degree, I volunteered to help out with informal computer classes for disabled kids at a nearby boys’ high school. Most of the other tutors were seniors from the school, supposedly intimately familiar with the school’s Apple II machines, which were what we were using. But they had only programmed in BASIC. It turned out they also had LOGO boot disks, which the boys had never used. I started up one, and gave a quick intro to the joys of list processing, doing quick-and-dirty word-for-word language translation from English to French with just a few lines of code. They were quite impressed, and agreed that the same sort of thing would be very hard to do in BASIC. And yes, LOGO was designed by Seymour Papert to be easy for kids to understand. And he has done and published the research with actual kids to prove it.
    5
  119. 3:41 Why did you need to pass the string source of the function? You could have passed the function itself. That’s perfectly valid Python.
    5
  120.  @profdaveb6384  Just don’t go taking part in any duels until then. ;)
    5
  121. It’s actually a “directed acyclic graph”. A commit can have a single parent, multiple parents (merge commit) or even no parents at all (like the first commit in a repo). Trivia question: can there be more than one commit in a repo with no parents?
    5
  122. 9:56 The trouble is that you end up stretching your definition of “ain’t broken”. Turns out it crashes if you press two keys in the wrong order? Well, you learn never to do that. Entering records with a certain value in a certain field will corrupt the database? Never enter that kind of data. So your entire business process becomes convoluted to avoid all the sharp edges. There’s a hidden cost to that, but you never even think of how that might compare to the not-so-hidden cost of replacing the system with a better one.
    5
  123. Aren’t there double-microphone designs, where the forward one directly picks up the speaker’s voice, while the one behind picks up the room reverb so it can be subtracted out? I recall Apple had something like this in its “PlainTalk” microphones from the early/mid 1990s. What has happened since then?
    5
  124. Around about 1990, back when hard drives were a lot smaller and people still used floppies a lot, JPEG was an absolute revelation. The idea that you could take a photorealistic image and reduce the file size by a factor of 10 with no perceptible difference in image quality was stunning.
    5
  125. 6:17 What’s missing is called “dynamic range”. Also note it’s not about large versus small numbers, but large versus small magnitudes of numbers. Remember that negative numbers are smaller than positives ones (and zero).
    5
  126.  @sundhaug92  Probably deleted by YouTube’s anti-spam bot. You can never be sure what it will take exception to. One time I couldn’t even post a link to another YouTube video.
    5
  127.  @NoNameAtAll2  Sounds like you have made up a circular definition to try to support your point. Otherwise, tell me where in the HTML spec it defines what “simplest” means.
    5
  128. Apropos a comment made by an ancient computer scientist about a different language, but perhaps applicable here: “it’s the language of the future, and always will be”.
    5
  129. You can do it in colour, too.
    4
  130. The paradox of cache-heavy modern architectures: RAM may be cheap, but accessing RAM is expensive.
    4
  131. The collision between “Plan To Throw One Away” and “Second-System Effect” was particularly amusing ...
    4
  132. 14:03 The ’286 was never (“nominally” or otherwise) a 32-bit chip.
    4
  133.  @timothycurnock9162  How does the key know who it has been given to?
    4
  134. Social science is considered a “soft” science. That’s a physicist/mathematician’s way of saying “not a real science”. Remember, this is not just any o ld “project management” -- this is about the management of software-development projects. If you don’t understand what’s so special about that, you need to read Brooks’ book.
    4
  135. Paging was a tricky thing. When we first installed MACSYMA on our VAX-11/780, it behaved very badly. Then our sysadmin edited the .EXE file, changing the page fault cluster size from its default of 16 down to 1, and that improved things a lot.
    4
  136. If you’ve read anything by Douglas Hofstadter -- particularly his classic Gödel, Escher, Bach: An Eternal Golden Braid -- that law of his starts to make more sense.
    4
  137. “Multi computer” would imply entirely separate physical boxes connected by cables. These might be regular LAN interconnects, but custom high-speed ones are also used (e.g. Fibre Channel, I think). One also talks about “tightly coupled” versus “loosely coupled” multiprocessor machines. Typically “loosely coupled” ones don’t share RAM, so they communicate by sending messages over communication channels of some sort. “Tightly coupled” ones share (almost) their entire RAM space, so they can communicate just by reading and writing shared memory. A small step back from full tight coupling would be NUMA (“Non-Uniform Memory Access”), where each processor has some local memory, and can also access memory belonging to other processors, but the latter accesses are slower. So this logically appears just like tight coupling, and is almost as easy to program.
    4
  138. 4:09 Python enforces indentation, too. But it allows you to use either tabs or spaces, just so long as you are consistent. Forcing you to use only tabs is a step backwards, in my view.
    4
  139.  @CTimmerman  Who might be running GUI frontends. Or they might not.
    4
  140. You know that modern browsers compile it to machine code, right?
    4
  141. IBM’s excuse for EBCDIC was that ASCII was almost, but not quite ready for prime-time in 1964.
    4
  142. That’s how interrupts are handled in CPUs anyway.
    4
  143. Those IBM mainframes seemed to have amazingly primitive filesystems. Even our little DEC PDP-11s had files that you could grow just by writing to them.
    4
  144.  @LiamMayfair  Modern operating systems don’t tie themselves to a particular GUI layer. Microsoft and Apple do, but Linux doesn’t.
    4
  145. There might be one in this box. Or there might not. You can’t tell until you open it.
    4
  146. Anybody who was around in the 1980s might remember Judge Greene’s “Modified Final Judgement” on the breakup of AT&T, the old US phone-company monopoly. Just to say, if lawyers and judges can do it, why shouldn’t software developers do it too?
    4
  147.  @djp_video  I have never heard of these sorts of problems from any other user of Git. It has a reputation for rock-solidity. Remember how it became popular: it did not have the backing of any big company, no management mandating its use, only a small cadre of smart open-source developers who used it to gain a quantum leap in productivity. No slick, multi-million-dollar advertising campaign: only the first-hand testimony of real users. And its adoption grew from there. But this is all beside the point. If it pains you so much, why do you insist on using it? Maybe you would be better off with something like Microsoft Team Foundation Server. Then at least you can blame Microsoft for your problems.
    4
  148. Mel seemed to manage it OK.
    4
  149. 2:47 The important point is that the same inputs always produce the same outputs.
    4
  150. 0:07 You mean “off by 2” if you said “a year early” instead of “a year late”!
    4
  151. A brand-new Linux installation only requires one reboot. Last I heard, Windows requires about five.
    4
  152. And then there’s the fact that Windows loses its “New-OS smell” after just a few months, and needs reinstalling to get it back. Linux manages to stay fresh for much longer.
    4
  153. The Alien vs Predator of data interchange formats!
    4
  154.  @cromptank  I agree, that would be a better basis for setting the origin of a calendar than some fictitious birth date of some deity.
    4
  155. 7:08 Tell your colleague that rust isn’t magnetic.
    4
  156.  @nextlifeonearth  Only 3?? Luxury! I just counted the occurrences of “if(UNIX AND NOT APPLE)” in the build scripts for the Blender source code, and found 24 of them. Oh, and one “if(UNIX AND NOT (APPLE OR HAIKU))”. Yet another almost-but-not-quite-Unix system ... sigh ...
    3
  157.  @myguitardidyermom212  Reread what I said: “linguists found little or no use for his mathematical innovation”.
    3
  158. I have that C++ book. Not as useful as the actual language reference.
    3
  159.  @busimagen  Whoosh right back atcha.
    3
  160.  @kc9scott  Emacs, on the other hand, can go full graphical. You can even have clickable buttons in the middle of a text buffer.
    3
  161. 0:14 Another world map without 🇳🇿 ... sigh ... 0:37 Mentioning that he wants to ping Auckland just doubles the irony ... 0:44 ... OK, that’s a better map!
    3
  162. 3:57 Notice the switch numbers up to 21? (The other end starts at 0.) The 11/70, while it was a 16-bit machine with a 16-bit “virtual” address space, had 22-bit physical addresses. It think it could handle up to 2MiB of RAM, not 4MiB--the other half of the address space was for accessing I/O devices. But the software couldn’t access the entire address space directly — it had to keep fiddling memory-mapping registers to map various 8kiB blocks of the virtual space to desired blocks of the physical space.
    3
  163. 6:16 But then, if you are always filling rectangular areas rather than plotting individual points, then you will always be specifying the bounds of the areas to be filled. Thus, filling the whole playing field would mean filling a rectangle with corners (0, 0), and (width, height), not (0, 0) and (width - 1, height - 1). In other words, stop thinking of pixels as little squares. You want squares, think in terms of squares, not pixels.
    3
  164. 3:58 LEDs always need a resistor to stop them exploding, because they are prone to an effect known as “thermal runaway”. Most substances exhibit higher electrical resistance as their temperature rises. So as current flows through them and heats them up, this increases the resistance and acts as an automatic limit on how much more current can flow. LEDs exhibit the opposite behaviour: their resistance goes down as they heat up. Thus, if you put too much current through them, they heat up, lowering their resistance, and this means even more current flowing through, until they blow up, die etc.
    3
  165. Pascal was such a (relatively) small language, there were loads of compilers available, including ones you had to pay for and ones that you got for free. ALGOL 68, on the other hand, was much larger and more complex, so in our little neck of the woods, we never saw a compiler for it.
    3
  166. 10:35 Another kind of SIMD was introduced back in in the 1970s with the Cray machines. These had instructions which could act, element-by-element, on arrays of up to 64 elements with a single invocation. However, the operations were not performed simultaneously on every element, but were pipelined so that, after the initial instruction setup, the result elements would be churned out in succession very quickly. Contrast this with PowerPC Altivec (which I think predates Intel’s initially rubbish MMX), where operations are performed on all elements of the operand vector simultaneously. Note also that these vectors tend to be shorter than the Cray ones: a maximum of 4 elements (maybe 8 for small integers) is typical.
    3
  167.  @BruceHoult  Ho yus! He was the one who violated the terms of the “Don’t Make Larry Unhappy” licence!
    3
  168. 2:20 Unfortunately not “real” Windows. So it’s not even a “real” version of a “not real” OS. ;)
    3
  169. 6:09 Actually that was done using fixed-point (scaled-integer) calculations. Which were quite a pain to work with, but saved on floating-point hardware. As the hardware cost dropped, eventually it got to the point where the saving from leaving floating point out was negligible compared to the cost of programming without it. This was about the early-to-mid-1990s.
    3
  170. I would have to say, [citation needed].
    3
  171. I wonder why we don’t say at some point “OK, memory is now big enough, we can afford to go back to static RAMs now”. For example, you can easily pack in something like 64GiB of DRAM in a modern machine. If the same amount of circuitry can only amount to 16GiB of SRAM, that would still be very useful for many purposes. Also. SRAMs are faster to access, which is why they are used for high-speed caches. And think of the power saving!
    3
  172. Trivia question: what’s the difference between Smalltalk metaclasses and Python metaclasses?
    3
  173. 6:20 There is another way to write a command-line program, and that is to get its inputs from command-line arguments. Options passed by keyword can typically be specified in any order. This is typically more flexible than asking the user to type input lines.
    3
  174. In GUIs (0, 0) is typically at the top-left, anything touched by 3D graphics tends to put it at the bottom-left. It’s one of those minor annoyances, like the ordering of keys on the numeric keypad on computer keyboards versus telephones. Stick in the appropriate y-inverting scaling matrix somewhere and then just forget about it.
    3
  175. This is the baby version. Grown-up adults use sheepda calculus.
    3
  176. Which perhaps is why mainstream development is moving away from Windows. And why Microsoft is trying so hard to turn it into Linux.
    3
  177. The mistake lies in thinking that there is a “left” and “right” in memory. There isn’t, any more than there is a “top” or “bottom”. There are only numbers for bytes/words, and numbers for bits within a word. Spatial arrangement only applies when a human looks at a diagram. The usual reading order for English (or any other language using some variation of the Roman, Greek or Cyrillic alphabets) is left-to-right, top-to-bottom. If the bit numbers within a word should be in decreasing order from left to right, then, for consistency, shouldn’t we have word addresses decreasing from top to bottom?
    3
  178. Unix was the model for Linux, which today totally dominates the computing world. Once it became commercialized, Unix unfortunately became the field for a tussling match between a whole bunch of different vendors each trying to lock customers into their own proprietary variant. This fragmentation ultimately destroyed Unix. But oddly enough, at the same time, Unix was also the core nursery for the Free Software/Open Source movement, among other things the GNU project for creating a Free operating system. This is where Linux began. So Unix was very much the launching pad for what became the heart and soul of the computing world today. Its spirit lives on.
    3
  179. Asterisk is fun. I set it up for a client quite a few years ago, and even did some custom programming for it. That was an audio-only application, to replace analog phone calls.
    3
  180. 9:00 “Sheer speed” of IBM’s hardware!? Somewhere, Seymour Cray is going “HAHAHAHAHAHAHA!!!”
    3
  181. Don’t worry, everybody says it. ;)
    3
  182. OS/360 (the troubles with which started off The Mythical Man Month ) was written in assembler, but that didn’t stop it being complicated, buggy and slow.
    3
  183.  @gregoryfenn1462  I think the OP is referring to the thought processes, not something superficial like the accent.
    3
  184. Your own local branches are quite separate from remote branches, and you can have any number of each. “git branch -r” show remote branches, while “git branch” shows local ones. Remote branches are copies of other people’s public branches.
    3
  185.  @quangtung2912  What is the relevance of “precomputed” to what I said,. exactly?
    3
  186. 3:18 Presumably there are additional checks somewhere to ensure the packet length is sane. Otherwise you could fiddle those first four bytes to some random bizarre values and cause mischief that way. Because remember the MAC at the end has to be found via the packet length, so you can’t verify that until you are sure of the packet length!
    3
  187. TLS elbow-bump FTW!
    3
  188.  @johnfrancisdoe1563  Ðat’s what I þought.
    3
  189. 10:05 Yes, you could move directories to other partitions. I wouldn’t advise doing it to important OS directories while that OS is running, though. This is why SystemRescueCD was created. Keep a USB stick handy with a copy of it around.
    3
  190. Warning: Using TFS or Visual SourceSafe can be seriously harmful to your programmers’ productivity.
    3
  191. 7:41 UCSD Pascal was very influential. Most of the PC-based implementations were clearly inspired by it (e.g. variable-length strings, “univ” pointers).
    3
  192. You should ask yourself: is your project as large and complex as the Linux kernel? Over 20 million lines of code, thousands of new and changed code flowing in each day from over a thousand active contributors, with a new major, stable release coming out about every three months. And they manage it all with Git. I’m guessing that your particular project is a lot smaller than that. So if they can be so productive with Git, and you cannot, guess where I suspect the problem lies?
    3
  193. 3:00 Consider that the Unix kernels were themselves written in C. So a userland C program calling printf would pass control through the runtime library into the kernel, which would then do ... what? It can’t be language builtins all the way down, at some point there has to be actual reusable code to implement I/O.
    3
  194. Can’t handle regextion? 😅
    3
  195. There is failyr standard lambda-calculus notation, e.g. f = λx.x But I agree, that Haskell notation doesn’t make any sense to someone who doesn’t know Haskell.
    3
  196. I bought a Mindstorms kit, but I never used any of the proprietary software. I found the open-source NBC/NXC compiler, and built my code with that. I also wrote some of my own tools (in Python) to encode/decode the custom sound and graphics formats.
    3
  197. The reason for that should be obvious: it’s because Emacs was originally doing “windowing” on text terminals, before GUIs became commonplace. Rather than change their terminology, they simply added another nonstandard term, “frames”.
    3
  198.  @AndersJackson  ISO 8601 FTW! Also the national standard in Japan. For all their quirks, that’s one thing they did that makes sense.
    3
  199. Where does he get all that tractor-feed paper from? There’s nobody still making it, is there?
    3
  200. It was probably efficient taking the cost of the hardware into account. As in, with the technology available at the time, there was no way to increase the utilization efficiency by much without spending disproportionate amounts of money.
    3
  201.  @AntonioPetrelli  The nice thing about an intermediate representation is that it greatly reduces the porting burden. Consider the problem of implement M different languages for N different processor architectures. If you had to write every compiler to directly generate code for every architecture, that means writing M × N different compilers. But if you write M different language front ends to generate IR, and N different back ends to translate IR to processor-specific machine code, then the work required is on the order of M + N.
    3
  202. Pai certainly lied about the “DDos attack” on the FCC site.
    3
  203. 5:31 Slight irony in IBM’s dominance, though: the most popular programming language for commercial applications was COBOL, which IBM wasn’t really keen on. It kept trying to push its customers to use its own “Commercial Translator”, but they kept pushing back and insisting on COBOL. So finally it had to give way. COBOL was also one of the first “portable” (or at least, largely portable) languages. So a nice side effect of its dominance was that it weakened vendors’ lock-in of their customers.
    3
  204. It’s like Java for the 2020s.
    3
  205. Remember the old analog computers? Back in the early 20th Century, analog computers could solve certain problems faster than digital machines, but with less precision. They fell out of favour as digital computers became able to match them in speed while maintaining greater precision. Also they were never any good at number-theoretic problems (e.g. factorization, code-cracking). Quantum computers are basically a reboot of the old analog-computer idea. The kinds of problems they can solve are in the same class -- physical simulations and optimization -- where speed is more important than precision. And they are similarly useless at number-theoretic problems.
    3
  206. 10:06 I would have used Inkscape for the graphics. It’s a better design tool.
    3
  207. 10:37 “Committee” or not, ALGOL 60 was actually considered a major advance on languages that had gone before. The spec was remarkably small, and only partly because it left out important stuff like I/O and if-then-else nesting rules ;). It had nice ideas like recursion and local variables (and of course call-by-name), which were sufficiently challenging to implement that they were a major driver of advances in compiler technology for some years to come.
    3
  208. 6:26 Not just an odd number, but a prime number: 7. One that cannot be evenly divided up into anything else! Before byte addressability became de rigueur , computers had word lengths like 24 bits, 36 bits, 60 bits -- all numbers with lots of integer divisors including powers of 2 and 3, even 5, so they could be divided up into equal-sized portions in many different ways. But with byte addressability, the basic unit is 8 bits, and the natural machine word length (so far) is 2, 4 or 8 times this. So the only factors you have are powers of 2, which limits the ways you can divide them up.
    3
  209. and you want to see how much space is being used just by the current source code, not including the version history. If you do du -ks .git . then it shows you the space used by the .git subdirectory, and then it subtracts that from the usage it shows for the “.” directory. Because it knows the earlier directory is contained within the later one!
    3
  210. Actually, every computer is “quantum”. Consider that the behaviour of the transistor (charges tunnelling across an energy barrier, “holes” (missing electrons) behaving just like positively-charged electrons) can only be explained in terms of quantum theory.
    3
  211. Spoken like a true telephone engineer. A retired one.
    3
  212. The Blitter was fast, but limited. It couldn’t do rounded-corner rectangles. The Macintosh UI, for example, had rounded-corner windows and rounded-corner buttons. On the Mac, it was all done in software, which was slower but more flexible.
    3
  213.  @mheermance  RISC was very much about turning traditional assumptions on their heads. That’s where a lot of speed came from.
    3
  214. 9:49 Basically, there are no “revisions” or “second editions” of RFCs. Instead, they get replaced with entirely new RFCs with different numbers.
    3
  215. 6:52 No, ASCII 1 actually (the “SOH” character). Off-by-one error... I think a NUL (ASCII 0) is done with CTRL-SHIFT-@.
    3
  216. pid content tracker”. He wouldn’t trust his code to the elaborate conflict-resolution schemes that some people were coming up with .So Git provides the tools to narrow things down to just the relevant places needing fixing up, with a minimum of irrelevant detail, and leaves you to sort it out.
    3
  217. Worth contrasting the telephone system, based on 19th century technology where you have to remember someone’s telephone number (or maintain a directory on your phone), versus the Internet, developed in the 20th century, where the network itself takes care of finding the numbers for you, you just have to remember their names. The mobile phone in your pocket is such an advanced piece of technology, yet when you make a call or send an SMS to someone, it still falls back to this 19th-century way of finding them through the network--by a number instead of a name.
    3
  218. Not really worth mentioning proprietary languages.
    3
  219. 7:50 It allows the shared library to be updated separately from your executable. For example, a newer version with newer features, bug fixes etc ... so long as it remains backward compatible, you don’t need to recompile or relink your executable to take advantage of it.
    3
  220. That’s still true in the Free Software world.
    3
  221. ALGOL-W was also known as “Wirth-Hoare ALGOL”. Think of it as a prototype of ideas that Wirth would later embody in Pascal.
    3
  222. Bitbucket started out as an online repo service only for Mercurial. It got forced into providing Git as an alternative, by popular demand. And then the popularity of Mercurial diminished to the point where it wasn’t worth supporting any more. So that is now gone.
    3
  223. It wasn’t Apple’s fault for coming up with a situation that C cannot handle gracefully.
    3
  224. As I recall, one reason why a lot of that simplicity had to go away was because shell scripts and subprocesses do not work well under Microsoft Windows. I think this was about the time Mozilla were looking to move to one of these newfangled distributed VCS things (likely they were using Subversion before). But one important criterion was that it had to work well under Windows. Sadly, that ruled out Git (at the time). So they went with Mercurial instead.
    3
  225.  @skipfred  You mean special method names like _add_ or __getitem__? They allow custom operator overloads much more simply than C++ can manage. Even Java, which was supposed to be simpler than C++, cannot offer such a feature.
    3
  226. 2:15 Originally, a “spline” was an actual rod made of some flexible material, that a draughtsperson would use to draw curves on paper with a pencil. Then the term was carried over to mathematical operations for doing the same sort of thing.
    3
  227.  @flyball1788  You don’t just want to be able to diff(1) them you also want to be able to patch(1) them and be able to do merges on them. If you have a tool that can convert them to/from a text format, why not store that text format in the repo, and convert it to binary as part of the build process?
    3
  228.  @HighestRank  Current macOS/OS X is officially “Unix”. Apple are a licensee of the trademark.
    2
  229. Who says open source cannot be “commercial”? Free software means “free of restrictions”, you can still make money from it.
    2
  230. ImageMagick and command-line tools in general are all about getting the computer to perform repetitive tasks, so you don’t have to.
    2
  231. It is also possible to stimulate the nerves into seeing colour intensities that do not correspond to any real light input.
    2
  232. “WPF In C# 2010” ... I sincerely hope that is not being used much any more ...
    2
  233.  @busimagen  Figured out what “whoosh” means, yet?
    2
  234. Yes, worth talking about Lanczos. Also worth talking about the general principle of image/signal reconstruction using the sinc function, because all resampling techniques are basically shortcuts to avoid the high computational expense of sinc.
    2
  235. 4:23 That’s what a compiler for a high-level language does. The original name for this concept, from the 1950s, was “automatic programming”. At that time, the high-level language source was not considered the “program”: that name was reserved for the actual machine code that the computer could interpret.
    2
  236.  @kc9scott  This is why I have an Emacs window always open. The scratch buffer is a handy place to compose long command lines before turning them loose on bash.
    2
  237. I wish the ALGOL line had been more influential. In particular, it introduced the “:=” operator for assignment, leaving “=” with its mathematical meaning of equality comparison. Whereas every language in current use has adopted the C convention of “=” for assignment, and this peculiar “==” thing for equality comparison. So tell me, if “==” is for equality, why is the inequality operator not “!==”?
    2
  238. 7:30 Back when reading and writing were first invented, no doubt there were similar discussions about the usefulness of that. Initially it was a specialist field in its own right (you learned to be qualified as a “scribe”). Then it became a universal skill. That was the first “information technology” -- it gave us the ability to keep track of large amounts of information, beyond the quantities we could keep in our heads. Computing is the second “information technology” -- it gives us the ability to process large amounts of information, beyond the capabilities built into our heads.
    2
  239. 11:39 That’s great, but you’ve completely defeated the point of multithreading in this example. Now only one thread is computing at a time.
    2
  240. How would you do it? With text files, you can narrow it down to lines which have changed, and show those lines to a human user. What would the equivalent be for binary files?
    2
  241.  @zxuiji  The problem with locking is it slows down development and makes things awkward, like when you find that somebody has locked something they didn’t need to, and you have to ask them to unlock it so you can work on it. That’s why locking was done away with. That’s why productivity is so much greater with modern VCSes than those older, locking-based ones.
    2
  242. The context for that comment was I think the phenomenon that people who have just discovered regexes tend to (over)use them for everything. One horrendous example was a regex for parsing e-mail addresses.
    2
  243. Those who remember BASIC also remember the immediacy of being able to type a statement and have it executed right away, versus sticking a line number in front of it to add it to the in-memory program. That was very convenient for experimenting, and for developing a program incrementally. You can do something analogous, I would say nicer, with Jupyter. You can type any number of lines of code (in Python or other supported languages) in a cell, and execute that cell with shift-enter. And so you can incrementally build up your code across multiple cells. And there is never a need to do the equivalent of a SAVE command, because the notebook system is regularly autosaving for you.
    2
  244. 7:21 fsck has nothing to do with checking for bad blocks. That’s only a filesystem consistency check. fsck still exists, though it needs to run less often now that journalling has become commonplace.
    2
  245. There is another problem with JPEG, and that is if you push the compression too hard even on images without sharp edges in them, it ends up looking blocky. There is another compression technique, known as “JPEG-2000”, that uses wavelets instead of DCT. Here when you push things too hard, it gets fuzzy rather than blocky, which many people feel is less objectionable.
    2
  246. I expect you charge suitably eye-watering sums to make up for such short-sightedness ... ;)
    2
  247. 19:49 Correct layout of mathematical equations is a solved problem, surely. Consider that even HTML-based Jupyter notebooks can include correctly-rendered maths formulas, thanks to MathJax.
    2
  248. You need the lamp. The Grue only eats people in the dark.
    2
  249. The older distribution upgrades will still be available, on some kind of archive site. So you should have that option available pretty much indefinitely.
    2
  250. 2:39 “Xilinx” is pronounced “zai-linx”, not “zillinx”.
    2
  251. Windows is not that popular with developers. Why do you think Microsoft is trying to add a Linux layer to it?
    2
  252. youtube-dl is still available.
    2
  253. If they were using Microsoft Team Foundation Server, you can understand why anything would be seen as an improvement on that ...
    2
  254. I use Linux. I have used a range of languages with Linux. Before I used Linux, I used a range of languages on a bunch of other platforms. I not only understand what a “curried function” is, I have used them in real-world situations. Would you like a “so what?” lecture? I could give you one.
    2
  255. Fun fact: Functional programming is older than OO.
    2
  256. @Michael Hansen I think the earlier ILLIAC IV research machine was also SIMD, with four processors executing the same instruction set on four data streams. Seymour Cray looked at the difficulty of programming that before creating his architecture for the Cray-1.
    2
  257. 8:24 The trouble with the hardware blitters in the Amiga and Atari ST is that they were never as powerful or flexible as software. The Macintosh had its QuickDraw graphics routines, which implemented sophisticated drawing modes and nonrectangular clipping. When you tried to build similar functionality on top of a more limited hardware blitter, the performance would often end up worse than if it was all done in software.
    2
  258. Modula-2.
    2
  259. I recall a colleague, a former IBMer, mentioning how some of those pages went beyond the usual “THIS PAGE INTENTIONALLY LEFT BLANK” business. Instead, they said “DESTROY THIS PAGE”.
    2
  260. If you look at the three different kinds of address numbering needed for memory, they are * byte addresses within memory (call this number B) * bit numbers within a byte for masking (call this number b) * bit numbers within a byte for integer digit values (call this number a) In all little-endian architectures, we have a = b and B = int(b / 8) which is very simple and straightforward. But in big-endian architectures, the situation is more complicated. Also, another peculiarity of big-endian architectures is that CPU registers are still effectively little-endian! For example, consider an instruction sequence like move_word a → b move_byte b → c Does c end up with the high or low byte of a? In little-endian architectures, it is always the low byte. In big-endian architectures, it is the high byte if b is in memory, but the low byte if b is a register! Thus, the only truly consistent byte/bit layout is little-endian.
    2
  261. 0:37 Blur ... focused ... heh-heh.
    2
  262. Phosphoric acid dissolves your teeth!
    2
  263. 4:31 Marketing travesty generator! Is there a copy of that text anywhere? ;)
    2
  264.  @RobertMilesAI  Fun fact: Python allows semicolons, so you can put multiple (simple) statements on one line! Its rule for omitting semicolons is a bit saner than the one in JavaScript.
    2
  265. Isn’t there such a thing as an echo-cancellation microphone? The front one picks up your voice, the back one picks up the room reverb, and by subtracting the latter from the former, you get a much clearer, less boomy recording.
    2
  266. But will they help him find the Golden Fleece?
    2
  267. No, it’s really the opposite of antialiasing.
    2
  268. 1:24 Someone who has no recollection of Life Before Linux ...
    2
  269. Luckily, there are Open Source alternatives that are much more modern, like R, Octave, and of course Jupyter notebooks using Python with Matplotlib.
    2
  270.  @TrueThanny  Yes, address books are precisely my point. Why do we have to do the lookup of the number, and then have to dial that number? Why not have the computer do the lookup for us? That’s how the DNS works -- it’s like an address book that automatically keeps itself up-to-date!
    2
  271.  @Richardincancale  I’m sure there were smart researchers at IBM, but somehow their shipping products rarely reflected that -- they were clunky, resource-hungry and overcomplicated compared to, say, DEC and Unix systems.
    2
  272.  @Harry-lu7hj  C# does it all for you ... on only one platform. “26 drive letters ought to be enough for anybody!”
    2
  273. Java doesn’t “ban” pointers, it only obscures them. As a consequence, it suffers from the aliasing problem, e.g. a = b; a.field = «value»; System.out.println(b.field); /* what does this print? */
    2
  274. One drawback with DNS A records, in particular, is that they only give you an IP address, not a port number. So they are not sufficient to identify a service, only a machine which might provide that service. This was remedied later with the introduction of SRV records, but they are not heavily used.
    2
  275. Interlacing should die already.
    2
  276.  @muhamednourezzeddin2784  The whole point with comments is that they’re not part of the “data structure” (or indeed any semantic content).
    2
  277.  @Carewolf  No, that’s not error diffusion. Error diffusion involves keeping track of the error between the true pixel value and what you have actually displayed, and accumulating (“diffusing”) that into subsequent pixels. Hence the name.
    2
  278.  @Carewolf  Floyd-Steinberg is one of many algorithms under the category of “error-diffusion”. It’s not the only one.
    2
  279. 0:39 Two places where it’s not used: SSH and VPN connections. They have their own encryption handshake protocols. Make that three: wi-fi also has its own security system(s). Bluetooth? Four. Among the places where it is not used ... nobody expects the encryption inquisition!
    2
  280. That’s called “key management”. It’s an entirely separate issue. Perhaps you do it by turning up in person at the registration centre with suitable ID.
    2
  281. There is a more fundamental aspect of the “laziness” principle not mentioned here, and that is that a computer is a _universal machine_, capable of being programmed to perform any kind of tedious and repetitive task you can describe to it. Consider the number of people who use GUI tools to perform repetitive tasks, for example, when a command-line tool could be scripted to perform the same thing automatically, leaving the human free for more interesting tasks. GUIs have their uses, but they cannot be automated.
    2
  282. 4:09 Ah, the joys of Baudot code ... Baudot gave his name to the unit “baud” ... which is another term you don’t hear used much any more ...
    2
  283. 2:02 The I/O problem mostly went away of its own accord. Once punch cards, paper tapes, magtapes, line printers and all the rest of it went away, everybody pretty much settled on something Unix/POSIX compatible (stdio). Writing to that gives you a nice, portable common denominator.
    2
  284. The Debian emacs-common package is just about 95MB.
    2
  285. I can’t even see anything in standard SQL that allows columns to be referenced by numbers, only names. Not in the SQL 2003 spec, anyway.
    2
  286. The private key never leaves your custody.
    2
  287. Sonic screwdriver!
    2
  288. One end still has to own both halves of each public/private-key pair.
    2
  289.  K.D.P. Ross  Because pure-functional languages tend to be awkward to use. Procedural languages which allow functional constructs (like Python does) tend to work better.
    2
  290. 1:34 Well, by “completely changed” you mean that somewhere around 50% of the bits in the hash flip to different values. You don’t want them all to flip (or flip according to any discernible pattern), otherwise that’s no longer quite so random.
    2
  291. No, C++ and Python do not let you define new operators, they only let you overload existing ones. ALGOL 68 did allow you to come up with new operators. But you couldn’t use non-alphabetic/numeric characters for them.
    2
  292. 6:09 Idea: why not let the robot loose with that idea, and see what solution it comes up with? This is how we give birth to the evil robot overlords, isn’t it?
    2
  293. Many implementations did “lazy buffering” to get around this. For example, DEC’s VMS compiler (which I liked because of its system-programming extensions).
    2
  294. 11:54 Yup, all the Free Software desktop environments can agree on common standards as prescribed by freedesktop.org, I think is the main place where these things are discussed. They’re not out to wrest control of the marketplace from each other. So your choice of DE doesn’t even have to dictate your choice of what apps you can run.
    2
  295. Macs are the ultimate example of how something can officially license the “Unix” name, yet work nothing like people expect a traditional “Unix” system to work.
    2
  296. Since you ask an off-topic question, let me give an off-topic answer: Have you tried turning it off and on again?
    2
  297. An 11/70 was the main campus machine when I started my Comp Sci degree (1979). I learnt a lot from messing around on it--stuff that wasn’t even taught in lectures. Several of the lecturers seemed to have expertise elsewhere. For example, one handed out notes on how to do system service calls on a UNIVAC machine, because that’s all he knew. Which was completely useless, given we had no access to such a machine. Nobody gave a lecture on how to do service calls on RSTS/E — what a wasted opportunity, when there was an actual living, breathing machine running that very OS just outside the lecture hall, that the students could have run actual exercises on.
    2
  298. Yes, they do talk to ordinary nobody clients like you or me. In fact, a lot of their traffic is completely unnecessary, coming from misconfigured clients that should be sending their queries elsewhere. But they can’t block it.
    2
  299. Ah, is that why the limit.
    2
  300. 7:30 Uncanny: I had a graduate CS lecturer who reported the exact same experience.
    2
  301. SSH client bug? How would the client even know which part of the output was from wall versus your program?
    2
  302. I think his screwup was in not properly configuring the change in volume layout.
    2
  303. Yes, it could very well happen again. People will get lazy again and forget what happened last time. Remember, it’s not a millennium bug, it’s a century bug.
    2
  304. Cloudflare also offers 1.1.1.1, in case you want to have an alternative to Google.
    2
  305. 3:12 By “letter” do you include everything in Unicode classes Ll and Lu? Because, you know, we don’t live in an ASCII-centric world any more...
    2
  306. What happens when your audio/video is being received by more than one other conference participant? Does multicasting work in practice, so a single copy of the data is transmitted from your machine, and received by multiple recipients? Or do you have to send out multiple copies, one to each recipient?
    2
  307.  @jursamaj  Now you begin to understand why Git became popular, in spite of the dominance of older, big-name products: it’s because it makes the merge problem actually bearable. With prior VCSes, merging was fraught with difficulties and opportunities to screw things up. With Git, it it actually becomes something you can perform as a routine task.
    2
  308. Triple meaning. “Current”!
    2
  309.  @stanrogers5613  I thought there was no Windows 9 because Germany said no.
    2
  310. the old US phone-company monopoly.
    2
  311. You can solve that the same way it’s done in TCP -- by putting sequence numbers on the packets.
    2
  312. 10:22 Methinks ’tis time to make a backup of the backup!
    2
  313. Which was possibly why they had no problems with publishing all their hardware schematics, and letting other companies make copies. Nowadays the company lawyers would be on you like a ton of bricks.
    2
  314. 8:16 Case in point: the Atari Falcon had a Motorola 56001 DSP, because in those days general-purpose CPUs were not fast enough to do processing of CD-quality sound data in real time. Nowadays we don’t see DSPs much in regular PCs (except maybe embedded inside a controller for some peripheral). But new, more compute-intensive tasks have become popular, which is why we now have programmable GPUs, for example. Will the wheel turn again? Yes, I think it is possible.
    2
  315. Non-rectangular clipping regions are your friend.
    2
  316. beta, eta, theta, zeta ... they all rhyme.
    2
  317. “...and that’s what the flashing light is for.”
    2
  318. That was one game I managed to complete. I drew up a map by hand as I went through it. Figuring out the “twisty little passages, all different” was fun ;). In the end I remember I could complete it in about 40 minutes.
    2
  319. That sounds like an OS protocol stack limitation.
    2
  320. Conway’s Law: every engineering endeavour reflects the organizational structure that produced it. CORBA was produced by a committee of representatives from Big, Important companies who thought they owned the computing industry. Need I say more?
    2
  321. Actually printf is an interesting case, because it shows up the drawbacks of trying to do everything as a library routine within the limited type system offered by C. The trouble is that printf needs to be polymorphic -- accept arguments of any mixture of different types -- which C really cannot handle. So it simply gives up and doesn’t bother checking the argument types. There is another approach, which is illustrated by the memory-allocation API in Modula-2. The language defines built-in “NEW” and “DISPOSE” routines, which are fully type-safe, because they are constructs known to the compiler. But the language itself doesn’t define any memory-allocation routines. Instead, these constructs are translated to lower-level calls to routines named “ALLOCATE” and “FREE”, which are type-unsafe, and not defined anywhere in the language. Instead, they have to be provided somewhere in the program, or imported from some library. This way you get the best of both worlds -- type-checking at the high level, but full versatility of implementation at the low level. Imagine if something similar could be done on a more general scale, for I/O and possibly for other things. C++ uses a similar idea for its memory-allocation mechanism, but unfortunately its “<<” and “>>” I/O operators are a bit rubbish. And its printf support is just as unsafe as in C. GCC tries to do some type-checking of printf format strings. But this cannot work if the string is dynamic, since that requires the argument type information to be available at run-time.
    2
  322.  @SimonClarkstone  It had a name: “Jensen’s Device”.
    2
  323. @A Rogue FBI Agent Alexei Sayle’s recipe for the Great British Night Out was 93 pints and a chicken vindaloo.
    2
  324. 6:32 Interlacing is completely unnecessary with modern digital video transmission.
    2
  325. And yet it persists!
    2
  326.  @stefanluginger3682  IDEs tend to be inflexible in their support for different build systems. In Open Source software, there are a great variety of build systems in use. That’s why I avoid IDEs, and just use a powerful editor like Emacs. That way, it can drive any build system that the project might be using.
    2
  327. “Neat, tidy pile of GOTOs” ... that has to be seen to be believed. Sounds like a “neat, tidy pile of worms”.
    2
  328. 9:12 DEC were able to do the same sort of thing, a little bit later, at a fraction of the cost: the PDP-11 was their first byte-addressable machine.
    2
  329.  @MatthewStinar  COBOL was designed to be highly portable. Hence the specification in the standard that arithmetic always use 18 decimal digits, regardless of the platform. The vendors who were putting together the standard decided that 18 digits was equally inconvenient for all of them!
    2
  330. Certainly the insides of the Enigma were considered a closely-guarded secret.
    2
  331. If nobody else could figure out how to fit equivalent functionality into the available resources, then Mel did absolutely the right thing.
    2
  332. In Python, functions are first-class objects. So yes, they can be part of the state of an object.
    2
  333. Customers still wanted them. Even given the close architectural relationship between VAX and PDP-11, there were still people who found it easier to stick to the 16-bit platform.
    2
  334. The next logical thing to talk about after this would be Porter-Duff operators.
    2
  335.  @Spongman  You may care, but that’s no guarantee it will survive. You know how capricious Microsoft can be: look at all the ideas they have promoted over time, only to abandon them. Dotnet Core is not their first attempt to replace Dotnet -- remember what happened to Silverlight?
    2
  336. By the way, fun tip about the “du” command: if you type something like du -ks . that shows you how much disk space is being used by the current directory and its contents. Suppose it contains a .git subdirectory,
    2
  337.  @llearch  I’m not sure I would put up with working at such a place.
    2
  338. The CDC 6600 was a supercomputer. It could do computations about 50 times faster than anything else around. CDC was actually set up to sell business machines. But they made the mistake of hiring Seymour Cray as their hardware guru, and he just wanted to build the world’s fastest computers. And the company had no real idea how to sell them.
    2
  339. So how is it we Internet users can afford to interconnect using lower-latency landlines, and the news networks cannot?
    2
  340. RSX (and RT-11) could do async interrupt-driven I/O, unlike RSTS where all I/O calls blocked the process until completion. Only having access to a RSTS machine, I felt jealous. Until VAX/VMS came along ...
    2
  341. Lack of a hardware-imposed stack is not necessarily a disadvantage. Even POWER/PowerPC does not dedicate any register as a stack pointer; that is purely an ABI convention. And that’s about as modern an architecture as they come.
    2
  342.  @mheermance  On PowerPC Apple Macs, R1 was dedicated as the stack pointer. I think this might have been IBM’s usage as well.
    2
  343.  @mheermance  There’s even a note in the PowerPC documentation that load-with-update (i.e. stack pop) might be slower than doing separate load and add-to-stack-pointer-register instructions.
    2
  344. 9:09 You’ve already said that the registers seen by the OS kernel are not the same as those seen in user mode, even though they have the same names. So presumably the OS has access to special instructions for accessing the user-mode registers, in order to do this save/restore (and no doubt other purposes like obtaining arguments and returning results for SWI calls)?
    2
  345. No, the registers have the same names, but kernel mode has its own separate registers. This avoids having to save/restore registers when switching modes.
    2
  346. Needing to do arithmetic to hundreds or thousands of digits.
    2
  347. It’s no coincidence that the sound is so similar to that made by telephone modems, since they are both trying to solve nearly identical problems, namely how to use a medium designed for audio to transfer digital data. The obvious differences being: * Modems can ask for a retransmit when they hit an error, whereas the cassette player cannot be stopped and rewound to replay a section under computer control (unlike the tape drives on the bigger computers of the time, for example). * But on the other hand, even cheap audiocassette machines would be capable of better frequency response than the 4kHz standard bandwidth for telephone lines. Another thing is, modem coding technology continued to progress a lot after the days of using audiocassettes for data, reaching a peak of a 33.6kb/s transfer rate in either direction before being eclipsed by broadband technologies like ADSL. If similar trellis-coding techniques were retrofitted onto cassette encoding, it should be possible to load a program of, say, 128kB size in a matter of seconds rather than minutes. Or would that take more processing power than an authentic 8-bit machine could manage? And another thing to look at might be the incorporation of robust error-correction codes, like that on CDs and DVDs. But again, maybe this would require too much additional hardware or processing power to fit within the authentic 8-bit experience ...
    2
  348.  @the_hanged_clown  Classical Latin “C” and greek “K” are both pronounced like “K” in English. For a long time, English speakers seem to have had some trouble pronouncing words from other languages and peoples, so they would “anglicize” them. Nowadays, since English speakers themselves are from so many other languages and peoples, you’d think this effect would have lessened somewhat.
    2
  349. I never understood that whole pronunciation preciousness thing. Knuth tried to claim that the “X” (eks) was actually “Χ” (chi), and that the “e” (ee) was actually subscript “Έ” (epsilon), and so on. But the Unix/Linux commands are written “tex” and “latex”, so why continue to pretend they should be pronounced otherwise?
    2
  350. “JOVIAL” = “Jules’ Own Version of the International Algorithmic Language”. That “International Algorithmic Language” being one of the names for ALGOL-58.
    2
  351. You must hate Unicode.
    2
  352. 4:16 But watch out for the security implications of this, if the code you generate was based on input from a potentially malicious user. Unfortunately, I think this is the fatal flaw in the whole concept. This is why it is better to build your own interpreter on top of the base language -- as you could, for example, in Python -- and then generate code for this interpreter from the data. That way you get more control over what exactly the code can do.
    2
  353. Conflict resolution is one area where Git is a major advance on what came before. Linus Torvalds himself describes Git as the “stu
    2
  354. 12:30 Copy-and-paste, he has heard of it.
    2
  355. 5:15 “No” is the correct answer. Actually the idea of “graphical” programming goes back decades -- an early example was ProGraph on the Macintosh, back in the 1990s, and they have never really been able to make progress beyond that. Besides the objection that bwk points out, another major problem is that you can’t do version control on graphical program blocks. How do you diff two branches, and how do you merge them? We can do that with text code. Look at the massive popularity of Git nowadays, and the major difference that has made to our productivity. Nobody wants to abandon that.
    2
  356. 4:31 Marketing travesty generator! Is there a copy of that text anywhere? ;)
    2
  357. Where did it touch you?
    2
  358. 5:11 What is the point of having all or part of the name “Python” in the name of a Python module? Don’t you already know you are using Python?
    2
  359. 5:47 A better example might be a page fault. Here the process is trying to access something that it thinks is in memory, but in fact is not. The access is valid, the operating system simply has to suspend the process until the appropriate virtual page has been loaded into some available piece of physical memory before resuming the process, whereupon it retries the instruction that caused the fault the previous time, which now can execute successfully.
    2
  360. I don’t. I still take multiple steps: “git add -u”, check with “git diff --cached”, and then commit. Better to be safe ...
    2
  361.  @LikelyToBeEatenByAGrue  Did you buy them?
    2
  362.  @LikelyToBeEatenByAGrue  Relevance to my point being?
    2
  363.  @LikelyToBeEatenByAGrue  So what were you asking me, again?
    2
  364. I wonder how he would have coped with multithreading?
    2
  365. All computers are already “quantum” computers. Transistors (and vacuum tubes before them) would not work without quantum effects.
    2
  366. It’s not clear what the future of Dontnet is these days. Didn’t Microsoft try to replace it with Silverlight? Then WinRT? Then UWP? And now (almost) full circle back to Dotnet Core? Proprietary platforms do not have a long shelf life.
    2
  367. 3:17 And the reasons why the NSA came out with SHA-1 to replace the earlier SHA-0 (or just plain “SHA”) were not revealed publicly. But the weaknesses in the original SHA were discovered independently a few years later. This was part of a sequence of evidence indicating that the gap between public, unclassified crypto technology and what the NSA has was narrowing, and may not be significant any more.
    2
  368.  @ItsN1key  Trying to equate regular expressions with non-regular grammars is, shall we say, mathematically disingenuous, is it not?
    2
  369. xev is a handy program for demonstrating this. ;)
    2
  370. rsync is so cool. You know Andrew Tridgell developed it for his PhD thesis?
    2
  371. The password was something you could set in the source code.
    2
  372. Neither -- it was all FORTRAN.
    2
  373. CVS tracked changes on a per-file basis. So if you refactored your code and moved a function from one file to another, that had to be recorded as two separate commits. The problem with this is that, in principle, you have a state of your repo in-between those two commits, where the code is in an inconsistent state. SVN improved on this by allowing the recording of changes to multiple files as a single commit.
    2
  374. 0:44 No, 32 possible codes. The number of different characters that the Baudot code could actually transmit was 52.
    2
  375. You mean you don’t post your .emacs on GitHub, where you can easily clone it onto any machine?
    2
  376. 10:30 I think in your case you just needed to add an entry to /etc/fstab so it knew where to mount /usr from, and you would probably have been all right. But why was the SSD partitioned into separate volumes in the first place?
    2
  377. The 6502 family (which includes the 6510) was a highly-respected chip in its day. And the Commodore 64 was a pretty respectable machine, at least in the early 1980s, too.
    2
  378. 11:44 Probably there was no clipping as such, it simply chose whether to render each polygon in its entirety or completely omit it.
    2
  379.  @skipfred  I think Python is great. I have written—and maintained—much larger programs than that.
    2
  380.  @petitio_principii  Most 1980s-vintage graphics chips were fairly simple-minded: their job was to convert the state of bits in video RAM to an analog video signal that could be fed to a CRT monitor. All the real graphics rendering was done by the CPU. The one notable exception was of course the Amiga.
    2
  381. Both! It’s a tag team of high-powered multiprocessors and low-powered ones.
    2
  382. Feel free to show us how it’s done.
    2
  383.  @hanelyp1  Shared-secret authentication does depend on only the right parties, and nobody else, sharing the secret. How else would it work?
    2
  384. I use .gitignore to ignore files that everybody is likely to see, like build products. I use .git/info/exclude to ignore stuff that only matters to me, like my private “test” subfolder.
    2
  385. 7:35 Presenter points finger at one of an array of a dozen nondescript-looking chips. Audience nods sagely.
    2
  386. Certainly text files are preferable to binary. A key point about version control is about being to diff two different states of the tree to see what has changed. With a text file, you can see changes line by line; with a binary file, you can’t.
    2
  387. I think so, too. The resemblance is more superficial than deep. And BNF is a more powerful notation than DTDs. DTDs are not very useful with XML anyway.
    1
  388.  @gloverelaxis  “There is no such thing as "objective" or "unbiased" when it comes to information presented by humans. Everything contains bias.” Is that an unbiased comment?
    1
  389. Blender modelling is amazing. Animation is amazing. Physical simulation is amazing. Renderer is so amazing, proprietary programs are copying it. Python scripting leaves all proprietary apps in the dust.
    1
  390. Apple once tried to get away with MD4.
    1
  391. “Down” as in “less”, because for example -3 < -2, and if you want to talk about small numbers, -1000000 is much smaller than zero.
    1
  392. John McCarthy, the father of LI
    1
  393. SP, relates how he tried to suggest to mathematicians that functional LI
    1
  394. SP notation would be a less clumsy way of reasoning about computability theory, rather than the traditional Turing machine model with a print head and an infinite tape.
    1
  395. Apparently that viewpoint was not popular at the time.
    1
  396. Did you know SVG is an application of XML?
    1
  397. The only successful current quantum computer designs seem to be more along the lines of analog computers, rather than digital ones.
    1
  398. On most filesystems, space for files has to be allocated in whole units of some minimum amount, commonly called the “cluster size” (1KB on your particular volume). So even if the data occupies just 1 byte, you still need to allocate one whole cluster to the file. Some filesystems add a concept called “tail packing”, where these small sequences of leftover bytes at the ends of different files are stored in a common shared cluster.
    1
  399. And B&W TV worked so well because the eye is more sensitive to brightness/lightness than it is to colour.
    1
  400. No other word processor had a “Reveal Codes” function because no other word processor used embedded codes for formatting.
    1
  401. Trouble is it’s only a slave OS that expects you to control it from a computer running “real” Windows. Whereas Linux for the π is a full OS that can host its own development tools and other apps.
    1
  402. After many years’ experience, most of us have come to the conclusion that lex/flex and yacc/bison are not really worth using. Yes, you can generate automatic parsers for full LR(k) grammars, but we usually find we can get by just fine with LL(1) grammars, which can be handled quite easily with recursive-descent parsers. And hand-coding parsers and lexers makes them faster anyway.
    1
  403. No, mayonnaise an island.
    1
  404. 0 is 0, not positive. If you are encoding a signal as a signed integer, you would typically ignore the extra negative value, so that the positive and negative limits are the same. Otherwise your signal would end up with a DC offset.
    1
  405. The original 128KiB Mac could have its RAM upgraded to 512KiB. On top of that, some imaginative third parties found ways to push it even further. The “Monster Mac” upgrade pushed the RAM to 2MiB. It also modified the ROM so the smiling Mac that you saw on a successful initiation of boot had little fangs. ;)
    1
  406. The original 128KiB Mac could have its RAM upgraded to 512KiB. On top of that, some imaginative third parties found ways to push it even further. The “Monster Mac” upgrade pushed the RAM to 2MiB. It also modified the ROM so the smiling Mac that you saw on a successful initiation of boot had little fangs. ;)
    1
  407.  @busimagen  Some people see everything in choler ...
    1
  408.  @busimagen  Yes, Fuad. That is the reason why it is funny.
    1
  409. 2:29 The anointed solution for separating structure from presentation turned out to be CSS. That, like HTML itself, continues to be a work in progress. But it gets more powerful over time. It can even do animations now.
    1
  410. There’s nothing special at the hardware level to distinguish modifier from non-modifier keys--this is just a software convention. For example, in Blender 2.79, you can hold the D key down to draw with the Grease Pencil, or the K key and drag to cut with the knife tool.
    1
  411.  @keinunvergebenesaliasgefunden  Regular Emacs does all that just fine.
    1
  412. The “s” in “maths” is not silent: mathS!
    1
  413.  @matteopascoli  Ho yes. Only needed in languages which buggered up their interpretation of “==”, but still ...
    1
  414. You mean DRAM is like fine wine, and actually improves with age...?
    1
  415. True, I don’t like having a “return” statement. But the trouble with allowing the function name as a variable on the RHS is: how do you express recursion? I prefer the alternative of having an explicit name for the result variable, e.g. function factorial(n : integer) → result : integer is begin result := 1; for i from 2 to n do result := result * i end (*factorial*)
    1
  416. I do that sort of thing all the time. In C you can conveniently write “for (;;)” for an endless loop.
    1
  417. Every major Linux distro will have packages available in the standard repos for common stuff like pygame. I think Elementary is a Debian, or more specifically Ubuntu, derivative, am I right? If so, “apt-cache search pygame” will show you the right package names.
    1
  418. 10:46 Just noticed, that’s not a widescreen monitor. Is that a 4:3 ratio? I have both an 8:5 and 5:4 monitor within arm’s reach, and no matter how I lean to the side, I can’t make either of them look quite like that shape.
    1
  419. If Git is not installed on your production machine, why are you putting Git repos there? You just want the production version of the code, nothing more.
    1
  420. Yeah, that’s what I thought. Python has full support for functions as first-class objects.
    1
  421. I would say “my condolences”, but DEC is gone, too...
    1
  422. 7:28 Look at all the 10.* addresses up to step 7. That’s a lot of ISP-internal routing going on!
    1
  423. Consider that byte-addressability was touted as a big advantage back in the days of ASCII and other 7/8-bit character sets. Now that Unicode is becoming pretty much universal, is that still so important?
    1
  424. 6:04 Graphical programming languages sound like a neat idea ... until you try to manage patches for your code and do version control.
    1
  425. What @sporqist  said! One end or the other has to initiate the broadcast so the other can find it. Better for it to be the access point than your mobile device. Otherwise your device will be broadcasting the name of your home network wherever you take it. There was this peculiar thing called the “Free Public Wi-Fi” bug, back in the days of Windows XP, as I recall; all these devices were going around trying to connect to a network with that name, and nobody could figure out why. As one device broadcast that name, others would pick it up and try to connect to it too, and so the “infection” would spread.
    1
  426. XHTML was a failure. It was basically HTML reformulated with the strictness of XML syntax, which nobody could be bothered to follow -- not for web pages, anyway.
    1
  427. Neither: for (...) { ... } /*for*/ In C++ code, I have written things like for ( vector<DocTemplate::FieldDef>::const_iterator ThisField = Context.CertTemplate->Fields.begin(); ThisField != Context.CertTemplate->Fields.end(); ++ThisField ) { ... } /*for*/
    1
  428. 4:14 They went on to create the Free University Compiler Kit. When Richard Stallman was just starting his GNU project, he asked about using their code to build his GNU C compiler. They somewhat facetiously replied that, while the University might be Free, the Compiler Kit certainly wasn’t. So Stallman had to go off and create his own codebase for GCC from scratch. And the rest, as they say, is history. Does anybody still remember the Free University Compiler Kit?
    1
  429. 15:44 Arthur C Clarke worked in radar, too, I recall. Did he work with Wilkes?
    1
  430. File locking is how version control used to work back in the 1980s. Because it was thought that, if two people simultaneously made changes to the same file, the world would end or something.
    1
  431. You sound like a Java programmer.
    1
  432. 4:10 It could decode and display GIF images in fractions of a second--fast enough to play movies. No other machine could manage that.
    1
  433. 13:52 In fact, DTDs are a bit of an anachronism with XML nowadays.
    1
  434. Other VCSes, like SVN, automatically commit all changes you’ve made. Git doesn’t, because you often work on multiple things at once, and might prefer to split them up into separate commits, with handy commands like “git add -p”. If you forget what you’ve changed, “git status” will remind you.
    1
  435. 15:36 I suppose CSS animation features are right out.
    1
  436. 1:49 What is this “yes-yes-aye” of which you speak? In my day, we called it “SCSI”, pronounced “scuzzy”.
    1
  437. It’s a regular (Type 3) grammar.
    1
  438. 1:00 “High frequencies” i.e. image-processing-speak for “sharp edges”.
    1
  439. Has anybody asked him what version-control system(s) he uses, if any?
    1
  440. How do you think Apple became a trillion-dollar company?
    1
  441. I just wish C had used ”:=” for assignment, reserving “=” for equality comparison. I think DMR admitted he used “=” for assignment and “==” for equality just to save one character in typing. Also the preprocessor is unbelievably crap. Other than that, yeah, it’s a remarkably versatile language.
    1
  442. 10:03 Didn’t stop them running a FUD campaign against CDC, though. “Just wait for the 360/90,” they kept telling prospective customers. “It’ll leave the CDC 6600 in the dust. When’s it coming out? Soon.” Only it was two years late and completely failed to deliver the promised performance.
    1
  443. The Copper chip was far more innovative than the Blitter. It was the Copper that allowed the multiple vertically-sliding “screens” with different graphics modes. They could only be partitioned vertically, because the Copper chip could only change mode between scan lines, not in the middle of a scan line. But being able to dynamically rebuild a display list was (in those days) much quicker than having to copy entire screenfuls of pixels around.
    1
  444. 3:26 So I take it he doesn’t agree with Dijkstra’s proclamation that “computer science is as much about programming as astronomy is about telescopes” (if I recall rightly). As if programming is somehow merely part of the flooring of the subject, rather than lying at its heart. (I don’t agree with Dijkstra either.)
    1
  445. No big surprise, considering its lineage.
    1
  446. And also more expensive. Proprietary OSes, and the proprietary server-side apps that run on them, tend to be licensed based on the number of CPU sockets in your machine. A special exemption was made for multiple CPUs in the same socket, just for the sake of keeping the customers happy, because such a thing was unknown in the 1990s.
    1
  447. I can remember writing a program on VAX/VMS (this was the 1980s) to implement “pushd” and “popd” commands (save the current directory to switch to another one, then restore where you were before). The initial executable was just 10 blocks (5kiB) in size as I recall, but it took an appreciable fraction of a second to run each time, which I felt was too slow. But the VMS linker gave you the option of linking against a static version of the system library, instead of the dynamic one. I tried this, and the executable size went up to 50 blocks (25kiB). But it launched a lot faster. These days, on Linux, I really couldn’t notice the difference.
    1
  448. Not really. The original PDP-11 Unix system didn’t have dynamic libraries, and that had to fit within a 64KiB address space. VAX/VMS, by contrast, had a huge address space available, and typically ran on systems with memory measured in MiB. But that had shared libraries right from the beginning. These were not common on Unix systems until a few years later.
    1
  449. Technically, “dithering” is any kind of variation around a true value. Thus, addition of noise would be best described as “noise-based dithering”.
    1
  450. The context for that comment was I think the phenomenon that people who have just discovered regexes tend to (over)use them for everything. One horrendous example was a regex for parsing e-mail addresses.
    1
  451. 7:00 Church: the lambda-calculus man! Never mind trying to work with functions of numbers, imagine the whole concept of computation reformulated so that numbers themselves are functions! Ever wondered where the term “lambda” came from for a function defined at point of use without a name? It’s all his fault!
    1
  452. You have to type “sudo”?? You mean you don’t have a root terminal session already doing somewhere? Me, I always keep a couple going, in among the two dozen or so ordinary-user terminal sessions.
    1
  453.  @PixelPhobiac  Those terminal windows have a different theme with a bright yellow background, to remind me to take extra care with everything I do there.
    1
  454. 4:04 You mean lines were numbered from 1, not 0? Wasn’t this code written in C?
    1
  455. 7:57 Unfortunately, it seems computer graphics people can’t decide whether y should increase up or down: in 3D graphics it usually increases upwards, but in on-screen interactive 2D graphics, it more commonly goes downwards. Sigh...
    1
  456. The genetic code is self-replicating, if you consider the rest of the cell as some kind of general-purpose manufacturing machine.
    1
  457. We already are using 64 bits.
    1
  458.  @Gummibrandur  It’s a lousy way to do image compression, though.
    1
  459.  @shawn2736  That’s not “dithering”. That’s “screening”.
    1
  460. Grosch’s Law probably comes into it at some point as well.
    1
  461. I see my comment got “held for review” because it had the world “o
    1
  462. ld” in it. I can never tell whether a comment is “held for review” or just deleted, so I will leave both versions up.
    1
  463.  @robertherndon4351  If you want to include that, call it more like 18.6 bits. Oh, and I never said it did.
    1
  464.  @FrankHarwald  That one was no coincidence.
    1
  465.  @FrankHarwald  No, the one I mentioned was indeed a (near) coincidence.
    1
  466. It’s called “homogeneous coordinates”. Sometimes the extra component isn’t a 1; a value of 0 can be used to represent a “point at infinity” (having a direction but no actual position).
    1
  467.  @aarondavis5386  It means that all linear transformations in n dimensions can be expressed uniformly as multiplications between vectors of (n + 1) components and matrices of (n + 1) × (n + 1) components (e.g. 4×4 matrices and 4-vectors for 3D graphics). If you try to use n-vectors and n×n matrices, then rotation, scaling etc are represented by multiplications, but translation (change of location) has to be done by addition.
    1
  468. 2 bytes per sample × 2 channels × 44100 Hz sampling rate = 176.4 KB/s = 1411.2 Kb/s
    1
  469. The original 32-Bit Color QuickDraw from 1988 or 1989 implemented high-quality, but slow, dithering. Then QuickTime added a faster pattern dither for playing back an approximation of full-colour movies on limited-colour displays.
    1
  470. 2:06 Or there would be, if your hardware implemented “n-key rollover”, which means it can distinguish every possible combination of keys being pressed at once. Not sure if current keyboard hardware actually supports that.
    1
  471. C was hardly the first. At that point we had over a decade of experience with building large systems -- for example, IBM had already created PL/I by this point, and that was used for implementing MULTICS, which Bell Labs participated in for a while before pulling out and creating their own UNIX. And there was an even earlier language called JOVIAL, quite popular among the US military, which had the COMPOOL concept -- an early form of system library -- for storing common reusable definitions.
    1
  472. They took large numbers of programmers, working for years, to develop.
    1
  473.  @eddievhfan1984  The server doesn’t need any tickets; only the client does.
    1
  474. A never gets that key. It only gets tickets encrypted with that key.
    1
  475. Possibly ALGOL 68. Type VOID in there has only a single value, called EMPTY. So it was actually permissible to return a VOID value (i.e. EMPTY) from an expression.
    1
  476. 5:54 But the trouble with C# is that you are then stuck on Windows. Not the best choice for portable code.
    1
  477.  @EwanMarshall  On the contrary, rounding happens all the time. With currency conversions and compound interest, it is unavoidable. There is no such thing as arbitrary-precision fractional arithmetic.
    1
  478. I used to use the tab key, but only in editors where you could customize it to go in steps of four columns. This still caused problems looking at the text in other programs which did not have such an option. So I switched to spaces. Now I know my formatting will look the same regardless of what program is displaying it. For backward compatibility, I still have Emacs configured to sniff the file on opening; if it detects tabs, then the tab key generates tabs in that buffer, otherwise pressing tab inserts spaces.
    1
  479. 1:25 The sprocket holes were also punched as the data holes were being punched. So the tape did not have to be manufactured with the sprocket holes (synchronization track) pre-punched.
    1
  480. The fr
    1
  481. dom aspect of HTML, HT
    1
  482. TP and the W
    1
  483. WW was not popular with the established big companies.
    1
  484. When they saw this thing becoming popular, they kept trying to come out with alternative proprietary products to draw users away from the open standard: think of Quark Immedia, and Microsoft’s own “Project Blackbird” (and what MSN was initially going to be), just for examples.
    1
  485. They all died because they simply could not offer the kind of connectivity that the WW
    1
  486. W and the Internet already had built-in.
    1
  487. DJVU format deals with this by storing searchable text objects which are not rendered, separate from the actual page rendering. I think PDF allows this also.
    1
  488. It’s clear one of the design goals for PDF was that it would be easy to print on PostScript printers -- basically you prefix a PostScript prelude to define those one- or two-character PDF operators as synonyms for their direct PostScript equivalents, then send the contents of the page as is, end with a “showpage”, and out it comes. It seems in the beginning they couldn’t quite decide whether PDF would be a text or a binary format. They soon realized, of course, that conversions between differing OS-specific line-ending conventions would corrupt the object table if a PDF file was treated as text, so they added a recommendation (which might still be in the spec) that PDF generators include some random binary garbage near the start of the file, to stop format sniffers from concluding that they were dealing with a text file.
    1
  489. 12:15 No checking for errors from the pthread_mutex_lock call?
    1
  490. What’s the closest thing we have to a standard on-screen graphics API? For 3D, it’s probably OpenGL. But Microsoft doesn’t seem to like OpenGL, it keeps trying to push people to use its proprietary Direct3D instead. And then Apple is now promoting its own Metal API, while the rest of the world is embracing Vulkan. For on-screen 2D, X11 has long been popular, but it mashes together windowing and graphics rendering functions which should really be separate. So now we have Wayland, which concentrates on on-screen compositing and doesn’t try to be a full graphics API, instead leaving that to other components, like Cairo. Fun fact: there has been an effort for some years to make an adaptation of Cairo a standard part of the C++ library. Unfortunately it seems they are leaving out all the text functions.
    1
  491. I would like to know more about its daddy CPL, which was never quite implemented. BCPL is like a cut-down version.
    1
  492. 10:05 With Linux, was that measured in a GUI window (which GUI?) or in a text console?
    1
  493. In other words, Active Directory is less secure than true Kerberos?
    1
  494. With Git, it’s “commit early and often”. 😎
    1
  495. Coming initially from the DEC minicomputer world, I immediately appreciated the one unique feature of the little PCs, namely that direct memory-addressable frame buffer. Trying to do on-screen drawing by sending data down a serial line simply could not compare for speed and interactivity. But those early PC OSes never implemented a decent abstraction layer to allow apps to have some independence from the underlying video/graphics hardware. Instead, they just tried to be cut-down versions of big-machine OSes. Opportunity lost, I thought.
    1
  496. That’s why Apple charges so much.
    1
  497.  @jondreauxlaing  In Python, both functions and classes are first-class objects.
    1
  498. The Amiga could never handle non-rectangular windows, though. The Mac could.
    1
  499. 12:43 The answer is stored in a set of flag bits known as the “condition codes”. Conditional instructions can change their behaviour depending on the settings of these flags.
    1
  500. 3:34 What about “buckling spring”, e.g. Cherry MX switches?
    1
  501. It doesn’t, though.
    1
  502. 0:14 Ah, the notorious “«n» -1” loop count for DBRA on the 68000 family! Did nobody know how to write the loop correctly?
    1
  503. I used MPW for many years, with a few different languages. I never understood why other programmers preferred those more limited IDE thingies ...
    1
  504. That’s called “remote attestation”. It’s impossible to trust a PC that is under the control of someone you don’t trust. Unless it has something like a TPM chip in it. And that gets controversial.
    1
  505. 6:48 The PostScript graphics model was only ever designed to put marks on paper. It didn’t work very well for interactive graphics on a display screen. Also the language, while cleverly designed to be implementable with reasonable efficiency on common hardware of the time, was pretty crap (e.g. no lexical binding).
    1
  506. What difference does OO make?
    1
  507.  @rasz  It was never just about Intel. Unix workstations used quite a wide variety of chips then -- MIPS, PowerPC, Alpha, SPARC, HP-PA etc.
    1
  508. 1:01 IBM doesn’t make PCs any more. I think you meant “Microsoft-compatible PC”. After all, Microsoft has been the arbiter of “compatibility” ever since Microsoft Flight Simulator became the original benchmark for this back in the mid-1980s.
    1
  509.  @jasondoe2596  That’s the reality of it, though. Nowadays we are seeing other types of PCs, ones that owe nothing to Microsoft, start to become more popular, the obvious example being the Raspberry π family.
    1
  510. 12:33 Best to avoid spawning a shell in your subprocess if you can avoid it: spawn the traceroute program directly.
    1
  511. 1) Overhead. 2) Special interpretation of shell characters introducing unwanted effects.
    1
  512.  @tamasdemjen4242  We have technologies like autocomplete. It’s like speed-dial for the 21st century.
    1
  513.  @johnfrancisdoe1563  Names can also be matched using heuristics like Soundex, to save typing.
    1
  514. In those days, it was believed you had to lock files when they were checked out, because if two people made changes to the same file, the world would end or something.
    1
  515.  @Harry-lu7hj  Maybe you don’t know what Python is. Its core language definition is a fraction of the size of the C# one, yet it is a more versatile and dynamic language.
    1
  516. 8:30 And you can also verify n[a], to guard against fraudulent responses.
    1
  517. Imagine being able to write a code-analysis tool that can detect bugs such as endless loops in your program. This theorem says that such a tool is, in general, impossible. Also in general, it means that code-analysis tools cannot completely figure out what a program does without actually executing it.
    1
  518. 6:36 “Everything starts with zero”, or in other words, “everything starts with nothing” ... hmm, could be something profound in that ...
    1
  519.  @autohmae  What ideas did JavaScript take from AppleScript, pray tell?
    1
  520.  @autohmae  I used AppleScript for about a decade from its introduction in 1993. The only “event system” it had during that time was AppleEvents, and JavaScript has never had (or needed) anything like that. AppleScript was slow and made simple things awkward. For example, taking a substring of a string would raise an error if the substring was empty. Any other language would return a zero-length substring, but no, AppleScript required you to check for that situation as a special case.
    1
  521.  @autohmae  I used HyperCard for many years, mostly as a GUI framework. I wrote collections of XCMDs and XFCNs for accessing low-level functionality in the Mac OS, like QuickDraw graphics (e.g. drawing colour graphics when HyperCard itself didn’t support colour), sound calls, speech synthesis and quite a few other things. It certainly popularized a variety of ideas, but I don’t think most of them were original. For example, are you thinking of the “on-xxx” handlers in HyperTalk? So that you could attach little sequences of code to buttons, text fields and even an entire card, to respond to user clicks and other events. I would say that idea goes back to Smalltalk in the 1970s. Not only did Xerox PARC pioneer the concept of the GUI, they also invented event-driven programming to go with it.
    1
  522.  @elimalinsky7069  But SRAM uses less power, hence it puts out less heat.
    1
  523.  @elimalinsky7069  SRAM needs less complex controllers than DRAM. That’s part of the point.
    1
  524. It doesn’t for me. What OS are you using?
    1
  525.    Look at GIMP, for example, for which GTK was originally developed; in filter settings dialogs, there is usually a “preview” button, which lets me see the effect of the settings on the image in the main window. I can move the dialog around, scroll around the image etc and have it update in real time until I do “OK” to confirm the filter application or “Cancel” to restore the previous state of the image.
    1
  526. IPv4 or IPv6 makes no difference.
    1
  527. profiles for computer monitors, printers etc.
    1
  528. All the go
    1
  529. od image-manipulation apps will understand ICC data attached to images.
    1
  530. As for printed images drifting over time, the logical fix to that is to include swatches of known reference colours somewhere in or attached to the image. Then your sensor can simply apply a correction so the reference colours match up with the expected measurements, and you know the rest of the image should follow suit.
    1
  531. Has that table off to the side, behind him, had one of its legs sawn off?
    1
  532. 12:48 Interesting that the issue of “digital” versus “analog” seems to have fallen by the wayside by this point. I wonder how long it will take the “quantum computing” folks to realize that they have reinvented “analog” computers ...
    1
  533. That has already been set up.
    1
  534. Maybe a more general one on the meaning of “technical debt” ...
    1
  535. Check out the o
    1
  536. ld classic book Mindstorms by Seymour Papert, first published in 1980. This goes into the rationale behind the LOGO programming language, and is full of examples of the fun that kids could have learning to program.
    1
  537. Back then, BASIC was the go-to language for budding hackers with their newfangled home computers, but LOGO had list processing, like LIS
    1
  538. P! And of course turtle graphics. It was also a clever way to introduce mathematical concepts that some might consider too “advanced” for such young children. For example, drawing a circle as a polygon with lots of sides, something like (can’t remember the exact LOGO syntax): to circle : repeat 360 : forward a small step turn right one degree and you are in fact getting close to the differential-geometry representation of a circle. *Trying to avoid triggering the YouTube autocensorbot, which behaves particularly oddly on this channel
    1
  539. 5:22 It’s just easier to say “two to the sixteen”.
    1
  540. Any app that wants to provide a bulk search/replace function can benefit from regexes. Because you often don’t want to replace the exact same string, you want to replace strings matching a pattern. Also you may want to exclude spurious substring matches, and a regex allows for that, too.
    1
  541. I hated the whole “IBM-compatible” resistance-is-futile-you-will-be-assimilated Borg. I was a Mac man, and I appreciated CPUs with a flat 32-bit address space (Motorola 68000 family, and of course the ARM). I thought Intel x86 with its clunky segmented addressing was crap. But they did finally bring out a decent chip in the 80386.
    1
  542. Because the “parallel worlds” idea doesn’t really help explain quantum theory.
    1
  543. Why parallel worlds is useless: because it just replaces a superposition of states with a superposition of alternative realities. What makes one “plainer” than the other (Which is what “explaining” is supposed to do)?
    1
  544. That’s not the same TLS.
    1
  545. I used Blender’s mathutils.
    1
  546. kdtree module in some Python code a few months ago. Made it very easy to do things like create matching faces on mirrored vertices, or create mirrored vertex groups, without having to do it all by hand.
    1
  547. What platforms does Processing run on, that don’t have Python available?
    1
  548. 14:28 Threads are separate execution entities that share a common process context. It’s not the only way to parallelize a program: there are also programs that run multiple separate processes in parallel. In fact, one might argue that threads are about the last form of parallelism you should resort to: look at other solutions first if you can, then fall back to threads if nothing else will do. This is because of their propensity for the most difficult sort of bugs: the ones that are timing-dependent, so attempts to track them down can often cause them to (temporarily) disappear.
    1
  549. They got a patent on the QuickDraw region structure, and wouldn’t let anybody else use it until it expired in about the 200x’s.
    1
  550. 2:46 Forgetting to mention its key characteristic: the box marked “STORE” is used for both instructions and data. And how do you tell whether a memory word contains an “instruction” or “data”? Actually, you can’t: it all depends on how it’s used!
    1
  551. Actually, Python won’t let that happen. It disallows inconsistent mixes of tabs and spaces.
    1
  552.  @abram7547  Sure, I just tried it. A simple four-line script I called “space_test”: if True : print("hi!") print("hi!") #end if (That’s four spaces at the start of the first print, a tab at the start of the second one). Then when I tried running it: ldo@theon:~> python3 space_test File "space_test", line 3 print("hi!") ^ TabError: inconsistent use of tabs and spaces in indentation So you see, Python does prevent that sort of thing.
    1
  553. Sure, if you want to talk about databases, we should start with relational algebra. This is the mathematics of relational databases, and can express things that standard SQL cannot. Because, you see, “relational” ⊃ “SQL”.
    1
  554. It’s the other way round. There are specific classes of programs that can be shown to halt, but this is undecidable in the general case.
    1
  555.  @tristunalekzander5608  The public/private key pair are mathematically related in a way that anything encrypted with one can be decrypted with the other, but anybody in possession of the public key cannot (feasibly) determine the corresponding private key. Thus, the public key can be freely redistributed, without compromising the private key.
    1
  556.  @tristunalekzander5608  You use them one way round for confidentiality, the other way round for authenticity (digital signatures). Being able to decrypt something with a given public key guarantees that it had to have been encrypted with the corresponding private key -- in other words, that it came from the only party in possession of that private key.
    1
  557. Is there an example of a quantum computer cracking this?
    1
  558.  @kerrynball2734  They didn’t invent GUIs.
    1
  559. How would you know it was the same password?
    1
  560. That seems to be done for Apple’s corporate control purposes, not to safeguard the user.
    1
  561. Other platforms can implement security controls without forcing developers to go through a central vetting authority.
    1
  562. Wonder why Microsoft is trying so hard to turn Windows into Linux, then ...
    1
  563. Also there is this thing called “VRAM” (“Video RAM”). From the CPU side, it can be read and written like normal DRAM. But on the other side, there is a read-only port with a shift register that is continually transmitting the bits sequentially from memory. From there they are fed to video circuitry to interpret them as colour and brightness data etc before being output to a display. The point with this design is that the sequential output has low contention with regular CPU access, i.e. neither slows the other down much.
    1
  564. In Python functions can take functions as arguments and return them as results. Actually, it’s better than that. Python functions can take classes as arguments, and return classes as results. In Python, everything is a first-class object, and that includes functions and classes.
    1
  565. Also very useful together with the “find” command, to do multi-file search. E.g. reporting all occurrences (file names and line numbers) of “collada” in the Blender source tree: find ~/projects/blender -type f -not -path ./.git/\* -exec grep -FiHn collada {} ';'
    1
  566. Real hardware is amazing ... for those who can afford the space, and the electricity bills.
    1
  567. There is a critique somewhere by Martin Richards written after his experience porting BCPL to the PDP-11. He was pretty scathing of its byte-addressability. BCPL originated before the era of byte-addressability, and did not seem to adapt well to that.
    1
  568. I am impressed with the language extensions added in Fortran-90 and Fortran-95 for parallelizing complex operations. Quite handy for running on supercomputers. I suppose in C or C++ you use library calls to achieve the same effect.
    1
  569. 7:04 Notice the only kind of recursion allowed is tail recursion—there is nothing after the B that occurs in the substitution for B itself. This can be implemented with simple iteration, without a stack. Otherwise, it would no longer be a Type 3 grammar.
    1
  570. 4:58 the term is BINARY -- except that seems to have developed other connotations nowadays ...
    1
  571. 7:49 “Tape me to your leader!”
    1
  572. Yes, that can certainly happen. It’s called a “power glitch”. The usual cure is to unplug from the mains, wait 5-10 seconds, then plug it back in.
    1
  573.  @TomStorey96  That’s only if the POST can run.
    1
  574. 0:37 Maybe use the term “secret-key” cryptography instead. “Symmetric” sounds too much like stream-xor, where encryption and decryption are the same algorithm. Just to be clear on terms: “Private” -- something that only you know, and nobody else. “Secret” -- something only known to those who are authorized to know it.
    1
  575. Code is just data to a lower-level processor. What is the lowest-level processor? Is it the laws of physics?
    1
  576. 3:54 Not necessary to declare “grid” global since there are no assignments to that variable.
    1
  577. SNOBOL was a language designed for string processing. It could do some pattern-matching, but not quite to the sophistication of full-on regular expressions. (That I recall from descriptions, anyway; I never encountered it myself.)
    1
  578. As a cryptographer, you would have learned at some point that encryption and digital signatures are two inseparable sides of the same coin.
    1
  579.  @MrFair  You are the one who is not understanding basic things about the maths.
    1
  580. 3:20 I don’t know where you get the idea that quantum computers are going to have better luck at cracking public/private-key encryption than secret-key encryption. Currently they have proven completely useless at doing anything encryption-related, or indeed anything number-theoretic, at all.
    1
  581. Quantum computing is based crucially on the assumption that, for a linear increase in processing elements, you can get an exponential increase in processing power. This seems to come from someone taking too seriously the “many-worlds” mumbo-jumbo (mis)interpretation of quantum mechanics. It’s essentially just “something for nothing” snake oil.
    1
  582.  @NyanSten  What has worked, exactly? Has a quantum computer actually performed any number-theoretic calculation?
    1
  583. YouTube just showed me a 24-minute ad--that’s over three times longer than the video I wanted to watch. Never mind--I just let it play with the tab muted, while I did something else. That way you still get the ad revenue, right?
    1
  584. Mostly. They did manage to copy some AT&T code, which was only removed after a lawsuit.
    1
  585. 3:18 Presumably there are additional checks somewhere to ensure the packet length is sane. Otherwise you could fiddle those first four bytes to some random bizarre values and cause mischief that way. Because remember the MAC at the end has to be found via the packet length, so you can’t verify that until you are sure of the packet length!
    1
  586. Python is standard on Linux systems, so they shouldn’t have any problem running your Python code. But then, your Python code might in turn depend on other libraries. If they are common enough to be standard packages in the main Linux distros, then fine. Otherwise it all gets a bit complicated.
    1
  587. The trouble with Windows-style installers is, what if two different things you want to install have some dependencies in common? Does one overwrite the other? Or do they share? Or do you end up with multiple copies? How do you update them? This is why Linux distros have package managers, to keep track of the dependencies automatically, on a systemwide basis.
    1
  588. Should two-foot thick logs make that much difference? Unless their moisture content is a bit high ...
    1
  589. Yup, that’s the right way to think of it. Say you have a function which evaluates a cubic Bézier curve through given control points p₀, p₁, p₂, p₃ at a point given by a parameter t. Instead of expressing it all as a single function p ← B(p₀, p₁, p₂, p₃, t) you split the function call into two stages p ← B(p₀, p₁, p₂, p₃)(t) so you specialize it once for particular control points B’ ← B(p₀, p₁, p₂, p₃) and then the specialized function can be evaluated repeatedly for the specified parameter values p ← B’(t)
    1
  590. 3:34 Ha! He used “procedural” as a synonym for “imperative”, not (as is the fashion among some hipsters) as an antonym for “object-oriented”. Object-orientation tends to be mostly procedural, after all.
    1
  591. But Ethernet was.
    1
  592. BCPL was really a kind of interim bootstrap language for implementing CPL (either “Combined Programming Language” or “Cambridge Plus London”, named for the two universities collaborating on the spec while each producing their own compiler). I can find very little detail online about the actual features of CPL. Think of it as an all-singing, all-dancing 1960s kitchen-sink programming language that would have gone up against PL/I, except with some slightly more advanced computer-science concepts behind it.
    1
  593. There is a potential issue where you add commits to your copy of a branch, and then try to push it to a common server, and that fails because somebody else has already pushed their own commits onto the head of that same branch.
    1
  594. You would store the structure that the binary data represented.
    1
  595.  @sharpfang  Encode the chunks.
    1
  596.  @sharpfang  Represent the entire structure of the file as JSON structure.
    1
  597.  @sharpfang  Luckily, that’s not an issue with PNG.
    1
  598.  @AnastasisGrammenos  Yes, but none of that could “accidentally” be generated by server-side output.
    1
  599. A distributed phone book that your phone knows how to look up, so you don’t have to. And that is maintained automatically by the network itself, so you don’t have to.
    1
  600. See 12:22
    1
  601. All keys involving A except the one for initially communicating with S would have to be short-term, generated on the fly.
    1
  602. That’s part of the setup of B.
    1
  603. All keys not involving A would be long-term.
    1
  604. 13-minute ad for a 9-minute item?!?
    1
  605. Would you trust mission-critical business functions to an OS that can only handle 26 drive letters?
    1
  606. Windows still has that limit.
    1
  607. But none of the software will recognize it.
    1
  608. Here’s an example of where it falls down: a piece of software wants to install into the C: drive. You try to set up a mount point to use additional space on another volume, and tell the installer to put its stuff there. But the installer only checks for free space at the top level of C:, decides there isn’t enough, and the installation fails. This kind of rubbish is sadly all too common in the Windows world.
    1
  609. 1:06 If the format can be decoded by open-source software (e.g. dcraw), then yes, I think the odds are high it will still be accessible 22 years from now.
    1
  610. 3:42 That was true of one of my major clients back when I was still a Linux noob, too. They were running SuSE across their main departmental machines, so that’s what I put on my home box. Then later we decided that Debian was a better choice for servers. But Debian “stable”, while rock-solid for production use by my clients, was just too boring for my tastes. So I decided to configure my personal installation to use the “testing” repositories. Only, due to a slight slipup, I ended up on “unstable” instead. And that’s where I’ve been ever since. (There were also dalliances with Gentoo and Mandrake along the way. But I won’t bore you with those...)
    1
  611.  @RealCadde  But even English uses diacritics. For example, “resumé” versus “resume”.
    1
  612.  @JanBruunAndersen  Be careful, though. Those sorts of things can be localization-dependent, which is not what we want here.
    1
  613. Legally speaking macOS is Unix, but practically speaking it’s not.
    1
  614. Yes you can, but you will run into precision and stability problems. I remember going to a talk by a visiting lecturer, not long after the Hubble Space Telescope went up. If you remember, there was a mistake in the grinding of the mirror, which meant that the pictures were of lower resolution than originally expected. This guy explained how they came up with a mathematical model of the lens imperfection, and developed an inverse transformation to improve the quality of the images. They were achieving some quite respectable results. A later Shuttle mission took out one of the modules and replaced it with COSTAR, which fixed the optics to compensate for the mirror problem.
    1
  615. The concept helps with code reuse, and hence code maintenance. The less code you have to write, the less work you have to do, and the less chance there is for bugs to get in.
    1
  616. 1) Avoid using Microsoft Windows.
    1
  617.  Dusty 99  If avoiding unnecessarily complex OSes is “lazy”, I don’t wanna be hardworking!
    1
  618.  @WujuStyler  So far the only ones insisting on using Windows are gamers.
    1
  619. Echoing acoustics are the new normal. New normal. New normal.
    1
  620. Ever did any VAX programming?
    1
  621.  @RupertReynolds1962  It had a machine instruction that took a table of coefficients and would compute a polynomial for a given argument.
    1
  622.  @RupertReynolds1962  There were quite a few complex instructions similar to that. They really went to town in terms of making assembly language look higher-level than FORTRAN in many ways. Much of it turned out to be pointless when it was discovered that these complex instructions were slower than doing the individual steps using a longer sequence of simpler instructions.
    1
  623. No, the POSIX type is “time_t”.
    1
  624.  @mateuszsp8ebc691  No. In 64-bit Linux it is 64 bits. from <bits/timesize.h>: #if defined _x86_64_ && defined _ILP32_ /* For x32, time is 64-bit even though word size is 32-bit. */ # define __TIMESIZE 64 #else /* For others, time size is word size. */ # define __TIMESIZE __WORDSIZE #endif from <bits/types.h>: #if __TIMESIZE == 64 # define __time64_t __time_t #else __STD_TYPE __TIME64_T_TYPE __time64_t; /* Seconds since the Epoch. */ #endif
    1
  625. 0:19 Norgate ... isn’t that in Sussex somewhere?
    1
  626.  @JochemKuijpers  Strictly speaking no, because ASCII is only a 7-bit code.
    1
  627. 5:18 Answer should be (1000000 × 1000001) ÷ 2 = 500000500000
    1
  628. Many languages have had multi-value return, too. Not to mention Python generators, which can yield an unbounded sequence of values.
    1
  629. Binding a function object to its environment is useful in any dynamic language.
    1
  630. Why do I keep reading that as “Luau”?
    1
  631. 3:51 When I were a stoont, we had blackboards which were wraparound flexible sheets on rollers. So you just pulled on them to roll on to a new section. Eventually you wrapped round again to the part you previously wrote on, of course, so you had to erase it anyway, but in the meantime it could be left visible as reference.
    1
  632. 4:22 Interesting artifact of the video deinterlacing? Suddenly the tape seems like it is shredded into transverse ribbons.
    1
  633. Anybody who was around in the 1980s might remember Judge Greene’s “Modified Final Judgement”
    1
  634. on the breakup of AT&T, the old US phone-company monopoly. Just to say, if lawyers and judges can do it, why shouldn’t software developers do it too?
    1
  635. on the breakup of AT&T, the old US phone-company monopoly.
    1
  636. on the breakup of AT&T,
    1
  637. the old US pho
    1
  638. the ol
    1
  639. d US phone-company monopoly. Just to say, if lawyers and judges can do it, why shouldn’t software developers do it too?
    1
  640.  @Valvex_  My original one did not appear, and I was making no assumption that it was going to appear. I will leave the other ones up.
    1
  641. Are you trying to use it on Microsoft Windows, by any chance?
    1
  642. 6:39 Do you know how to use the du command to track down this excessive disk usage? There’s a trick involving bind mounts on Linux if you have your installation across multiple disks/partitions (as I and a lot of people do), if you only want to check directories on the root filesystem.
    1
  643. Shame about the MATLAB...
    1
  644. Single-level storage was pioneered in the ATLAS project, about two decades earlier. Also a feature of MULTICS.
    1
  645. My first language was FORTRAN. No, it’s not my favourite. Possibly because just a few months later I found out about POP-2, which totally blew my mind.
    1
  646. No, that does actually allow you to continue after pressing a key.
    1
  647. Think of it as implementing a higher-level abstract machine on top of a lower-level one. Only the lower-level machine is generating the higher-level abstraction on the fly.
    1
  648. Do you program for fun? Do it anyway. ;)
    1
  649.  @666Tomato666  I remember seeing a factor of 3000 times, so you’re not too far off. Also remember you’re not comparing like with like: hardware acceleration for AES versus none for RSA.
    1
  650. 0.21 You know what irks me? The “=” sign on his T-shirt in what is clearly a numerical approximation.
    1
  651. It’s a concurrent version of the sieve, where the array is one of processes. I have a version of it done in bash somewhere.
    1
  652.  @gestaltengine6369  How would it work? You’d have to give an example.
    1
  653.  @gestaltengine6369  But “multiple instruction” means the instructions can be entirely different.
    1
  654. That’s like saying that in the reals you can’t divide.
    1
  655.  @aDifferentJT  That’s like saying that in the reals there is no corresponding function that represents division.
    1
  656.  @fllthdcrb  Total nonsense, in other words.
    1
  657. What do you use instead?
    1
  658. What I am currently doing is checking after 60 seconds if the comment is still there.
    1
  659. C++ templates?
    1
  660. To go with complex numbers, Python has the cmath module, which is the equivalent of the math module but for the complex domain. Python tends to be my first resort for many things right now -- it has about all the high-level features you could want (e.g. functions and classes as first-class objects), short of turning completely into LI
    1
  661. SP. I agree using indentation for block structuring turns out to be a mistake;
    1
  662. I mitigate it a bit by having “#
    1
  663. end” comments to mark the ends of compound statements.
    1
  664. I also have custom commands in Emacs to jump between lines with matching indentation, which nicely lets me navigate between the matching beginnings and endings of compound statements.
    1
  665. People who appreciate science tend not to be susceptible to Poe’s Law.
    1
  666. 5:15 And wherever GCC goes, Linux follows ... !
    1
  667.  @AgentSquirrelA  Emacs can even edit non-text files. I successfully used it to apply patches to some proprietary binaries.
    1
  668. 6:08 And now in some households emergency services are only available via Internet. So why shouldn’t universal service provisions apply to that?
    1
  669. If you don’t know about TFS ... you don’t want to know.
    1
  670. Actually, I think they all do. Wasn’t DSA designed as an alternative to RSA that was only useful for signatures, not encryption? And didn’t somebody discover that you could use it for encryption, albeit very clumsily and inefficiently?
    1
  671.  @PauxloE  Any maths that can work one way, can work the other way. It may not be efficient or practical, but it’s possible.
    1
  672. Particularly when you combine it with something like screen or tmux at the other end. It just illustrates what a pathetic excuse for a command line Windows has.
    1
  673. Nobody should be using such insecure technologies any more.
    1
  674. I/O was a really difficult thing back in the day, with such a range of completely different devices with different behaviours--punched cards, punched paper tapes, line printers, magnetic tapes, disks and drums with entirely different filesystems. ALGOL 68 tried to solve the problem comprehensively, and ended up spending about a third of the reference manual just talking about I/O. It was really only the popularity of Unix, with its unified I/O model that vendors of other OSes started emulating, that made the problem finally go away.
    1
  675. 2:23 I think an early draft of ALGOL-68 very nearly violated that tradition by introducing “::=” as a syntactic symbol. Thankfully this was changed later.
    1
  676. 4:10 No, you are filtering out the high frequencies -- that is what a low-pass filter does. This leaves only the average of the on-off mixture, which is a greyscale.
    1
  677. Boomer is OK!
    1
  678. Holland is in the Netherlands, isn’t it?
    1
  679. 1:54 One problem with the Windows approach was that every GUI element (every button, menu, possibly even every menu item) was a “window” as far as the OS was concerned. And storage for every “window” was allocated from an area known as the “system resource heap”, which only had a small, finite size. So your machine could have lots of RAM, but if the GUI was too complex (or you had too many of them running at once), you could still run out of system resources!
    1
  680. As I understand it, this is still a problem.
    1
  681.  @kc9scott  That is correct. The genius of Steve Wozniak was in coming up with such a hardware-minimal design. And not just that, but one where the software work was done on such a low-powered processor. Engineers would look at the Apple II motherboard and ask “where’s the floppy controller?”. There wasn’t one. Or rather, it was the “RWTS” routine in the ROM.
    1
  682. How many classes does it take to implement a regex in Java?
    1
  683.  @mikaelkjericsson  Glad to hear it!
    1
  684. 1:55 What constitutes “detailed nuts and bolts” seems to vary over time, though. When higher-level languages were first invented, to take away the requirement to write in assembler, they were referred to as “automatic programming”, since the “program” was considered to be the actual executable machine code, the source code being a kind of “specification” of how to generate that. HTML still requires you to worry about the “detailed nuts and bolts” of setting up a form and defining its fields and their CSS properties, for example. If you want to go higher level than that, you resort to something like JQuery.
    1
  685. GUIs are handy for running lots of terminal and editor windows, and being able to copy and paste between them. 🐧🐧🐧👍
    1
  686. By the way, the “F” in “PDF” stands for “Format”. There’s no point in my sending you a “Portable Document Format”, but there is in my sending you a file in Portable Document Format. In other words, I don’t send you a PDF, I send you a PDF file.
    1
  687. That’s one of many different error-diffusion algorithms.
    1
  688. What’s the alternative? You’re not suggesting multithreading the GUI, are you?
    1
  689.  @PaulaJBean  You’re not suggesting we use callbacks instead?
    1
  690.  @PaulaJBean  Callbacks require fragmenting your UI logic into discontiguous pieces. This makes it harder to figure out what’s going on.
    1
  691.  @PaulaJBean  Who said it did?
    1
  692. Sounds like they copied the idea from Burroughs.
    1
  693. I used to read SIGPLAN Notices (the journal of the ACM Special Interest Group on Programming Languages) back when I was doing my Comp Sci degree. It was one of the, shall we say, more opinionated and less formal of the supposedly scholarly journals. I remember one little article suggesting that the ALGOL 68 convention of reversing the spelling of opening-bracket keywords to form the corresponding closing-bracket keywords “should be carried out to the letter, so to speak” -- namely, reversing the actual letters themselves!
    1
  694. At gigahertz clock speeds, it takes a long time for signals to get anywhere.
    1
  695. 0:18 A book on “Al-Qaeda” to one side, and “Teenagers” on the other. Which is scarier...
    1
  696. Windows developer, still believing that x86 is the (rapidly-shrinking) centre of the world? Compare an open-source OS like Linux, which runs on every major processor architecture for which GCC can generate code — that’s about two dozen of them at the moment. And userland apps that run under Linux are just as portable. Addition of new architectures hasn’t stopped. Consider the new hotness that is RISC-V. Meanwhile, Windows is still struggling to move off x86 onto ARM ...
    1
  697. Dependency issues are solved by integrating package management into the OS. That’s how we do things on Linux.
    1
  698.  @cmdlp4178  I think LLVM, like GCC is available on every platform on which the Linux kernel runs. That should be about two dozen major processor architectures. No other equivalent would even come close to those two.
    1
  699. GUIs are great for running multiple terminal windows. Copy and paste works nicely for command lines, which is something you cannot do with GUI operation sequences.
    1
  700.  @lincolnsand5127  Yeah, but copy and paste doesn’t work between them.
    1
  701. 5:58 Technically, “atomic operations”.
    1
  702. Somebody trying to downplay the use of modern languages in safety-critical applications might like to bear in mind that the touchscreen UI in the SpaceX Dragon capsule (you know, the one that just carried two crew to the ISS) is implemented using the Chromium web browser and JavaScript.
    1
  703. Why, what do you think it demonstrates?
    1
  704. You don’t consider the controls that the astronauts use to operate the spacecraft important?
    1
  705.  @jasonschuler2256  That is the primary way, the way they designed it.
    1
  706. 8:18 Of course, the logical next step would have been bit -addressability. A bit impractical with only 32 address bits, but now we have 64 bits, why has no one considered this worth doing? All transfers between CPU and RAM nowadays are actually in units of cache blocks, with smaller transfers happening only between registers and L1 cache, so the speed impact should be minimal. And it would simplify various kinds of bit-manipulation algorithms.
    1
  707. 0:04 At a guess I’d say that’s Python code on the right half of the screen.
    1
  708. Instinctive physics-undergrad answer: represent each point in 6D, being 3 spatial coordinates and 3 momentum coordinates. Then do ICP in that space?
    1
  709. What the cluck are you talking about? Oh, I thought you said “fowl” ...
    1
  710. I don’t like pure functional-programming languages, but I like using functional constructs in procedural languages like Python which make them very handy.
    1
  711. Linux is now the go-to platform to learn about computing. Microsoft and Apple are where you go to learn to press the right buttons and not ask questions.
    1
  712. 3:06 “Don’t want to like MATLAB” ... I can relate to that. Originally it was a FORTRAN-based program for manipulating matrices. But later on they tried to tack on other functionality to it. Earlier this century I had the misfortune of having to do GUI programming in MATLAB. Let’s just say I gritted my teeth and took the money ...
    1
  713. Makefiles on their own aren’t even enough these days, you need another layer on top, like CMake. For example, Blender is built with CMake. It also used to offer the option of building with SCons, but having two alternative build systems in the source tree seemed like a needless complication -- particularly when they got out of sync.
    1
  714. 5:50 That’s a job for auto-indentation, which is quite a separate function: press Enter, and automatically have the same whitespace inserted at the start of the new line that is at the start of the previous line.
    1
  715. Trouble with FORTH is, people seem to spend more time coming up with their own FORTH implementations than actually using it to write useful programs.
    1
  716. Maybe you mean “complexity”?
    1
  717.  @sugarfrosted2005  In that case, I have no idea what you mean by “relative computability”. Either something is computable or it’s not.
    1
  718. It must be a pain to have to put up with drive letters and .BAT files in this day and age. I’m glad I don’t have to use an antiquated toy OS.
    1
  719. Windows? I rest my case.
    1
  720. Macs can run Linux, too. In fact, every popular processor architecture on Earth can run Linux. Maybe you specifically mean “Microsoft-compatible PC”?
    1
  721. Microsoft’s ARM OSes are a joke. Apple? Apart from the tightly-locked-down IOS, it’s all vapourware. Or, judging from Catalina, likely to be just as tightly locked down.
    1
  722. Richard or David?
    1
  723. 11:28 Ahem Unchecked variants, anybody?
    1
  724. Lanczos or GTFO.
    1
  725. We’ve heard of one-hit wonders, but Linus really knocked it for six not once, but twice: first with the Linux kernel, and then with Git.
    1
  726. Anyone else notice you can’t mention Git
    1
  727. Hub in the comments?
    1
  728. Java bytecode, like Microsoft IL, is primarily designed for interpretation, with compilation more of an afterthought. LLVM is designed for compilation to efficient target code, not interpretation. It is built around SSA specifically for this reason.
    1
  729.  @imacds  I don’t see that at all. Small programs compile quite quickly on a Linux system. A large program, like Blender (more than a million source lines) I can build in about 10 minutes.
    1
  730. 19:58 That openness isn’t there today either, according to some companies. Proprietariness is alive and well. But at least Free/Open Source software has a name (OK, two names), and a clear definition you can point to.
    1
  731. I don’t know why that “keyboard not found, press a key to continue” thing is considered such a joke. It makes sense, after all, that if the hardware requires a keyboard to continue, it should wait for you to connect one.
    1
  732. The 68K architecture derives in many ways from the DEC PDP-11. I have programmed both. ;)
    1
  733.  @ScottyDMcom  PDP-11 was not 32-bit, though. The genius of 68K was that even the first 68000 was more than just a 16-bit processor, it was more a cut-down 32-bit design. VAX was arguably nice than both. But that’s another story...
    1
  734. Those were block-mode terminals. Proper GUIs can do things like change parts of the window in response to input elsewhere (e.g. enable/disable one option based on the setting of another, update a live preview image etc), while those terminals could not do any such thing.
    1
  735. But on Windows, it will likely crash before it manages to prove anything.
    1
  736. OK, here’s another one: how wide do you set your editor windows? I currently use about 100 columns, and sometimes I think I need more.
    1
  737. 8:03 Why is secret-key (what he calls “symmetric key”) encryption preferred to to key-pair (public/private) encryption for the bulk of confidential communications, instead of using the latter all the time? Because key-pair encryption is on the order of 1000 times slower than secret-key encryption, that’s why.
    1
  738. They really get up your nose.
    1
  739. If you look at the original spec, it distinguishes between no less than three representations: the “reference language”, the “publication language” and the “implementation language”. There was a recognition that typographic limitations on computer input/output devices would require conventions like quotes for representing things like begin and end in the implementation language, while the publication language (as used for examples in the spec) would write these as words in bold. In the “reference language” (the language as theoretically defined in the spec), these are not reserved words as such, they are symbols , and their actual implementation representation is not defined.
    1
  740. 5:05 No error checking on the pthread calls! Bad example.
    1
  741. Do you mean an audio file or a MIDI file?
    1
  742. 4:04 You don’t encrypt the packet length!? Even I was able to figure out how to do that. ;)
    1
  743. 9:45 YouTube’s comment system does not have durability.
    1
  744. I like to use the term “Dimdows” for the Microsoft product. Everybody knows what I mean. ;)
    1
  745. 1:19 JPEG does have a lossless option, I believe. Or there is always PNG.
    1
  746. Where do you think Robin Hood lived? Sherwood Desert?
    1
  747. Leap-seconds are a whole separate issue, necessitated by the slowing down of the Earth’s rotation. There is no predictable system to how they come in--it has to be determined by actual measurement. One thing we’re sure of, their frequency will inevitably increase over time. Some people are in favour of getting rid of them. But that leads to its own controversies ...
    1
  748. I use KDE Konsole, which lets you have tabbed windows. And I also like to use different themes for them. Notice he has a root session permanently open (upper left), for performing installations and other admin tasks? I put a distinctive yellow background on mine, just to remind myself to be careful what I type there.
    1
  749. One of the things unfortunately left unspecified in the ALGOL 60 spec was the meaning of something like if cond1 then if cond2 then alt1 else alt2 which “if” did the “else” pair with? Later languages specifically said that it paired with the “last unpaired “if”". But ALGOL 60 did not. Some implementations might have interpreted it that way, but others said you had to be explicit by putting in begin/end as appropriate, i.e. if cond1 then begin if cond2 then alt1 else alt2 end or  if cond1 then begin if cond2 then alt1 end else alt2 Of course, “begin/end” was only allowed in statements, not expressions, so this didn’t help with conditional expressions.
    1
  750.  @iii-ei5cv  Yes, but which John Steed sidekick? Tara King? Emma Peel? Cathy Gale? Or one of the older ones?
    1
  751. Feel free to come up with a solution.
    1
  752.  @LudwigvanBeethoven2  MouseMove needs to be carefully controlled. For example, the WaitNextEvent() call he describes takes a mouse-region argument, and a mouse-move event is only generated when the mouse leaves this region.
    1
  753.  @GilesBathgate  CIL/MSIL could not be used to implement its own interpreter, so you still need something lower-level than that. LLVM-IR is designed for this.
    1
  754. DEC’s 12-bit (PDP-5, PDP-8, PDP-12) and 18-bit (PDP-1, 4, 7, 9 etc) machines used this trick. The jump-to-subroutine instruction was called “JMS”, so JMS SUBR would store the address of the following instruction at location SUBR, and transfer control to address SUBR + 1. (cont’d)
    1
  755. (cont’d) By the way, the first prototype of what became Unix was developed on a PDP-7. You can see why they were quick to move to a PDP-11, which had a much nicer architecture with a proper stack.
    1
  756. (cont’d) This would jump to the address stored at SUBR, which was of course the point in the caller at which to resume execution. (cont’d)
    1
  757. (cont’d) There was no return-from-subroutine instruction as such, instead you ended execution of SUBR with an indirect jump: (cont’d)
    1
  758. (cont’d) JuMP I SUBR (cont’d)
    1
  759. And so I just discovered that if I used the right mnemonic for the JuMP instruction, YouTube deletes my comment!
    1
  760. And the latter is a modern superscalar RISC architecture underlying some of the most powerful computers in the world right now.
    1
  761. ... register defined by hardware, the IBM POWER/PowerPC architecture does not.
    1
  762. It’s no big trick to implement a stack in software.
    1
  763. ... had a dedicated stack-pointer ...
    1
  764. ... PDP-11 ...
    1
  765. While the ...
    1
  766. Now it seems I cannot post a comment with “dlo” (backwards) in it.
    1
  767. Remember that Torvalds himself describes Git as the “stu
    1
  768. Yes, there was a common ABI on VMS, which even included exception handling. So a language could throw an exception which got handled in another language! (Assuming your language notion of “exception” bore some resemblance to the VMS notion...) Nowadays it seems a platform ABI is defined for C and C++, and other languages have to conform to that.
    1
  769. 4:05 Another issue is the handling of autorepeat: by having this entirely in software, you don’t actually have to insert the repeated keystrokes into any buffers; you simply generate them on demand, as the user program asks for keystrokes. That way, you never have them happening faster than the program can handle.
    1
  770. 3:10 In ALGOL 60, a procedure body was defined as a single «statement», preceded by zero or more formal argument specifications. Of course, a begin/end block counts as a «statement», but it is not required to be this. This was later deemed to be too confusing, so in Pascal it must specifically be a begin/end block. Not sure if that was also the rule in ALGOL-W (aka “Wirth-Hoare ALGOL”), which was an intermediate language between ALGOL 60 and Pascal.
    1
  771. Best is to avoid such situations, by keeping your REs simple.
    1
  772.  @TofuBug24  That’s the kind of (mis)use of regexes Jamie Zawinski had in mind when he made his famous quote.
    1
  773. All it did was modify an index register that could only point at one of a set of labels.
    1
  774.  @TheOwlman  Well, that’s a problem with ICL, not the COBOL spec.
    1
  775.  @TheOwlman  In general that’s true of any variable: when it’s value is changed, there’s no “visible logic” to indicate it no longer has the value it was given at initialization time. So nothing too special about COBOL’s ALTER statement: it’s just a specialized form of assignment.
    1
  776.  @TheOwlman  It’s not as well structured as a case-statement, it is true. But given that you cannot assign arbitrary expressions in an ALTER statement, it is still possible to gather all the possible values of the switch-variable simply by finding all the places where it is mentioned.
    1
  777.  @fixups6536  It’s a switch statement. Quite a common feature in languages today, albeit in a more structured form.
    1
  778. Yes, that was a very influential language -- well worth a look.
    1
  779. There is a word which is used a lot in this video, which you are not allowed to mention in the comments. Can you guess what it is?
    1
  780. 11:18 Hilbert who, apparently, was quite miffed when Kurt Gödel came up with his proof of how no sufficiently-powerful theory of mathematics could ever be both complete and consistent.
    1
  781. Why always with the Australia digs? Not sure why I care ... I’m in NZ ...
    1
  782. Yup, hand-coding state machines to handle parsing various kinds of data is an extremely common thing. Done it lots myself.
    1
  783. Yup, hand-coding state machines to handle parsing various kinds of data is an extremely common thing. Done it lots myself.
    1
  784. The trouble with FORTH is it’s such a crummy language model. Even PostScript is better than that.
    1
  785.  @aleksandersabak  You like PostScript? Both its graphics model and the language are antiquated, but both can be improved.
    1
  786.  @aleksandersabak  No idea what you mean by “concatenative”. If you want to understand the difference between FORTH and PostScript, start by contemplating you forth love if honk then versus you postscript love {honk} if
    1
  787. User of Microsoft KlunkyOS?
    1
  788. Greetings from user [216, 25].
    1
  789. 3:54 I started out using tabs, too. But I used a reconfigurable editor which let you specify the width of a tab. So setting tab stops every 4 columns worked out nicely. Then, after many years, I realized that forcing everyone to reconfigure their editing/viewing tools away from the default of tabs every 8 columns was getting a bit wearying. So I switched to explicit spaces. Nowadays a I use an editor (Emacs) that can automatically sniff every file it opens. So if the file has tabs, then the TAB key inserts tabs. But if there aren’t any tabs, then the TAB key is automatically set up to insert spaces for that file. Also, having keystrokes to quickly adjust the indentation of blocks of text is very helpful.
    1
  790. 4:47 I think the spec was a bit vague about whether you actually had to specify the types of the arguments (integer, in this case). Some compilers may have allowed this to be optional. Which meant that type information had to be carried around at runtime.
    1
  791. Funny, then, that the world is running out of IPv4 numbers (addresses), but there are still plenty of domain names to go around.
    1
  792.  @fryersoncaptain  Not a very Linux-compatible Unix, though.
    1
  793. A password is a shared secret.
    1
  794. More likely you have/had /etc/nsswitch.conf, which lets you configure multiple sources for DNS (and other) data.
    1
  795.  @KuraIthys  But the price for SRAM is so high mainly because so little of it is manufactured. If it were more popular, the price would go down.
    1
  796. “Quantum” computers cannot even perform encryption yet, don’t expect them to handle any decryption.
    1
  797. You can also have different people/departments having access to the two.
    1
  798. Feel free to show us your PHP version.
    1
  799. Linux runs on everything. ;)
    1
  800. You may be right in comparing it to Java, given Oracle’s campaign against Google. But frankly, that’s not saying much.
    1
  801. 12:48 C-language trivia question: are the parentheses in “(1 << 24) - 1” needed?
    1
  802. Nobody has mentioned that he’s not just using Python, but Jupyter as well? 👍👍👍 Yay for Jupyter notebooks! 👍👍👍
    1
  803. This is why Linux systems have a package manager, to automatically keep track of these dependencies on a systemwide basis. E.g. apt-cache rdepends libcairo2 | wc -l 1557 That’s right, that’s 1557 other packages that depend on the Cairo graphics shareable library on my Debian system. Yet I only need one copy installed, and it only needs to be updated once.
    1
  804. But the digital graphics adapters driving them certainly did. And still do.
    1
  805. Perl is like quantum theory: if you think you understand it, it’s because it hasn’t been explained properly.
    1
  806.  @NoNameAtAll2  It can’t work, because opening/closing tags can be nested inside other opening/closing tags, to arbitrary levels. Regexes cannot handle that.
    1
  807.  @NoNameAtAll2  It can’t work, because opening/closing tags can be nested inside other opening/closing tags, to arbitrary levels. Regexes cannot handle that.
    1
  808.  @MSStuckwisch  Those are not regexes any more. Remember, regexes have to recognize a “regular grammar”. That’s a very specific technical term, referring to Type 3 in the Chomsky hierarchy, and nothing else.
    1
  809.  @MSStuckwisch  If you go beyond Type 3 grammars, then you lose their useful properties, such as the equivalence between non-deterministic and deterministic expressions, and the ability for them to run without a stack.
    1
  810. 2:49 Because then the language gets bloody huge . This was the approach taken with BASIC back in the days of 8-bit micros -- every vendor’s implementation had built-in commands to do screen graphics, disk file I/O, play sound, and every other feature they could think of sticking in. It just made the language too complicated.
    1
  811. OK, DOSser.
    1
  812. But what was the largest program JvN was ever able to write? In other words, how much absolute binary could he cope with, until even his legendary brain cells got full?
    1
  813. Memory is one thing; debugging a complex piece of code--keeping track of all the potential interactions between its parts--is quite another.
    1
  814.  @LittleRainGames  I’ve been thinking about this. “Conservatives” as supposed to be those who like to stick to how things were in the past. But they tend to exhibit very limited and selective memory of how things really were in the past. Those who actually study the history of how things were in the past, and try to learn from it, tend to get branded “Liberals”.
    1
  815.  @jasonschuler2256  Looks like some people don’t understand what “quantum” means. It’s not a marketing term, you know.
    1
  816.  @JNCressey  You mean like how a charge carrier in a transistor can be in a superposition of states on both sides of the gate? Which is how it can pass through the energy barrier and carry a (classically-impossible) current?
    1
  817.  @JNCressey  Except that the data doesn’t actually exist in superposition. That’s why “quantum” computers have such a high noise level, making them unusable for digital computations.
    1
  818.  @stromboli183  Cracking encryption will require fast number-theoretic computations. Which “quantum” computers have proved incapable of.
    1
  819. 10:46 Just noticed, that’s not a widescreen monitor. Is that a 4:3 ratio? I have both an 8:5 and 5:4 monitor within arm’s reach, and no matter how I lean to the side, I can’t make either of them look quite like that shape.
    1
  820. 12:03 It looks like a scan that has been quantized into a bilevel (black and white only, no greys) bitmap. Those little hairy extensions on the edges are characteristic of that.
    1
  821. You can do both encryption and signing of communications, to provide both privacy and authenticity.
    1
  822. 9:55 Airlines may offer different classes of seats at different prices, but airports treat all airlines equally. Imagine if airports were to charge you extra to board certain airlines and less for others? How would you feel then? That is the more accurate analogy: the airports are the ISPs, the airlines are the application services.
    1
  823.  MichaelKingsfordGray  None of them are in the same category as a decent platform API.
    1
  824.  MichaelKingsfordGray  You were trying to suggest that DotNet should be taken seriously, when even Microsoft is no longer so keen on it.
    1
  825.  MichaelKingsfordGray  “Gross blunder” hahaha. Consider why Microsoft is trying so desperately to make Windows more like Linux, and why they are now trying to push Dotnet Core which is open-source and cross-platform. Because the developers are deserting Windows for Linux. Windows-specific APIs don’t cut the mustard any more.
    1
  826. Leaving the device unplugged for 5-10 seconds isn’t just for the RAM, but for all the circuitry in the device to lose any stored charge. Unplugging and replugging the device too quickly can lead to glitches, where circuitry gets into a crazy state and the device locks up. This can happen during power brownouts, too.
    1
  827.  @TomStorey96  Memory tests have to be implemented by firmware/software, which cannot run properly if the hardware has glitched.
    1
  828.  @TomStorey96  Then you would know that leaving things unplugged is not a “carryover from a bygone era”.
    1
  829. That’s not a regular grammar.
    1
  830. Eclipsing binaries? Named for “al-gul”, aka “the demon”.
    1
  831. SHA-256 is just 32 bytes, or 64 hex digits. How many characters do you need to put in a tweet?
    1
  832. Yup. And in fact, the dnsmasq name server, which is designed for small setups like a home office, serves up exactly the contents of your /etc/hosts file, it doesn’t need (or understand) complex zone files like bind does.
    1
  833. Also known as the “Free University Compiler Kit”. Richard Stallman wanted to use that as the basis for GCC, but soon discovered that their idea of “Free” wasn’t “Free” at all. So today we have GCC, and nobody uses ... that other thing, any more.
    1
  834. The idea of an instruction set implementable across such a wide range of hardware capability was groundbreaking. The software side, not so much. IBM never really got used to the idea of timesharing (not until it adopted Unix in the form of AIX, anyway). Much nimbler competitors like DEC, DG et al ran rings around it in this area, during the 1960s and 1970s.
    1
  835.  @Richardincancale  CP/M was originally cross-developed from a DEC PDP-10 machine running TOPS-10. That’s where its inspiration came from, not IBM mainframes.
    1
  836.  @Richardincancale  The idea of
    1
  837. put
    1
  838. pu
    1
  839. tting a
    1
  840. do
    1
  841. t between the filename and the extension was not,
    1
  842. I think, something done in any IBM mainframe OS. Whereas it was characteristic of DEC operating systems.
    1
  843. Pu
    1
  844. tting a colon on the end of the device name was another DEC-ism. Only DEC systems had multicharacter device names; pity CP/M (and Microsoft) never copied that ...
    1
  845. The irony is that using vector graphics and actual text objects make it easier to edit the PDF file. The hardest type to edit is the one where every page is a bitmap.
    1
  846. 2:43 Not just any old paper, but tractor-feed paper!
    1
  847. C99 lets you use the words “and”, “or” and “not”. They’re defined in the standard include file iso646.h.
    1
  848. Nothing of interest happened in 1901 that you might want to record on a computer??
    1
  849. Well, that’s one way of dodging my question.
    1
  850. No it wouldn’t. Or, to be more precise, it might work in other situations (e.g. audio), but not here.
    1
  851. 1:07 It used to be “procedural” versus “functional”. Until the OO folks hijacked “procedural” to try to mean something different -- not realizing that most OO languages are in fact “procedural”.
    1
  852. 0:53 Not HTML, CSS, JavaScript/DOM, SVG, PNG ... no IETF RFCs for any of those.
    1
  853. 5:30 Lenovo ... this man has taste.
    1
  854. 4:04 You don’t encrypt the packet length!? Even I was able to figure out how to do that. ;)
    1
  855. “Call by name” is effectively “nameless function of no arguments”. Nowadays the idea of “lambdas” has become very trendy.
    1
  856. Apology accepted for thinking so.
    1
  857. export PAGER='less -iXx4'
    1
  858. But Debian would not do that. Not by default, anyway.
    1
  859. This is why I put blank lines, bracketing symbols and demarcation comments in my Python code, e.g. ‌ def outer_func ... : ‌ def inner_func ... : ... #end inner_func ‌ #begin outer_func ... #end outer_func ‌ #+ # Mainline #- ‌ ... outer_func(...) ... ‌
    1
  860.  @Lttlemoi  In Python, it’s just another method in the class. In C++, it’s a whole special function syntax, separate from the class.
    1
  861.  @Lttlemoi  Python lets you define them on subclasses of builtin types like int and str as well. You see, there are no PODs in Python. How’s that for consistency?
    1
  862.  @Lttlemoi  Python also lets you overload built-in functions like len() and abs(). Not to mention property getters/setters—have those made it into C++ yet?
    1
  863. Java was supposed to be simpler than C++, leaving out stuff they didn’t think was necessary. Then they discovered they needed some of that stuff after all, like generics. Or they try limping along without it, like the awkwardness of not having typedefs or unsigned integers. And so the reference manual ends up approaching the complexity of C++. And it still manages to leave out major things like operator overloads.
    1
  864. Try solving this without recursion: computing all the permutations of a sequence. def permute(seq) : "generator which yields successive permutations of the elements of seq." if len(seq) == 0 : yield () else : for i in range(0, len(seq)) : for rest in permute(seq[:i] + seq[i + 1:]) : yield (seq[i],) + rest #end for #end for #end if #end permute >>> list(permute((1, 2, 3, 4))) [(1, 2, 3, 4), (1, 2, 4, 3), (1, 3, 2, 4), (1, 3, 4, 2), (1, 4, 2, 3), (1, 4, 3, 2), (2, 1, 3, 4), (2, 1, 4, 3), (2, 3, 1, 4), (2, 3, 4, 1), (2, 4, 1, 3), (2, 4, 3, 1), (3, 1, 2, 4), (3, 1, 4, 2), (3, 2, 1, 4), (3, 2, 4, 1), (3, 4, 1, 2), (3, 4, 2, 1), (4, 1, 2, 3), (4, 1, 3, 2), (4, 2, 1, 3), (4, 2, 3, 1), (4, 3, 1, 2), (4, 3, 2, 1)]
    1
  865. Typical expensive and proprietary IBM technology. Also had its own reliability problems. They looked down on Ethernet, but Ethernet eventually wiped them out.
    1
  866. 15:34 Ah, but it could make a difference if you are trying to do very long baseline interferometry. For example, setting up a future radio telescope array. Maybe call it the “Square Light-Year Array”.
    1
  867. It is easier to be destructive than constructive.
    1
  868. Was NRS the equivalent of DNS SRV records rather than A records, then?
    1
  869. No, libc is in /lib, not /usr/lib.
    1
  870. 3:16 “If you can only take one programming language to a desert island” ... but wouldn’t your machine also have to be able to interpret build commands? In other words, wouldn’t it also need to have a shell? And isn’t that a programming language too? In short, one programming language on its own wouldn’t be able to achieve much these days.
    1
  871. Nevertheless, they are sufficiently related that what works for one can be made to work for the other.
    1
  872. Just FYI, checking the cert on this YouTube page, Firefox reports the encryption setup as “TLS_AES_128_GCM_SHA256, 128 bit keys, TLS 1.3”.
    1
  873. 16:00 I can’t imagine videos would work on readers with e-ink screens, though.
    1
  874. 7:06 Film was expensive. With digital cameras and camera phones, it’s easy (and cheap) to just go snap-snap-snap-snap-snap.
    1