00:00:00 --- log: started forth/18.10.15 00:03:25 --- join: smokeink (~smokeink@42-200-119-209.static.imsbiz.com) joined #forth 00:21:35 --- join: ncv (~neceve@2a02:c7d:c5c9:a900:6eaf:6ef7:3b81:d5f6) joined #forth 00:21:35 --- quit: ncv (Changing host) 00:21:35 --- join: ncv (~neceve@unaffiliated/neceve) joined #forth 00:33:18 --- join: wa5qjh (~quassel@110.54.188.224) joined #forth 00:33:18 --- quit: wa5qjh (Changing host) 00:33:18 --- join: wa5qjh (~quassel@freebsd/user/wa5qjh) joined #forth 00:48:46 --- quit: carc (Quit: QUIT) 00:50:16 --- join: carc (~carc@unaffiliated/carc) joined #forth 00:54:29 --- join: dave9 (~dave@90.20.215.218.dyn.iprimus.net.au) joined #forth 00:57:35 re 01:25:23 --- join: smokeink2 (2ac877d1@gateway/web/cgi-irc/kiwiirc.com/ip.42.200.119.209) joined #forth 01:25:31 how to disable this warning ? "DUMP is a system word in an application word in file" 01:26:45 --- quit: smokeink2 (Client Quit) 02:03:10 --- quit: ashirase (Ping timeout: 252 seconds) 02:06:58 --- join: ashirase (~ashirase@modemcable098.166-22-96.mc.videotron.ca) joined #forth 02:11:19 --- quit: dave9 (Quit: dave's not here) 02:19:24 --- join: smokeink2 (2ac877d1@gateway/web/cgi-irc/kiwiirc.com/ip.42.200.119.209) joined #forth 02:20:33 --- quit: wa5qjh (Read error: Connection reset by peer) 02:21:08 --- join: wa5qjh (~quassel@110.54.188.224) joined #forth 02:21:08 --- quit: wa5qjh (Changing host) 02:21:08 --- join: wa5qjh (~quassel@freebsd/user/wa5qjh) joined #forth 02:27:06 --- quit: rdrop-exit (Quit: rdrop-exit) 02:34:44 smokeink2: grorth? 02:34:57 I think you can just add "no warnings" to the top of the file 02:55:44 win32forth . tried adding "warning off" at the top of the file , it suppressed word redefinition warnings, but this DUMP warning still shows up 03:41:59 --- quit: wa5qjh (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) 03:44:36 --- join: wa5qjh (~quassel@110.54.188.224) joined #forth 03:44:36 --- quit: wa5qjh (Changing host) 03:44:36 --- join: wa5qjh (~quassel@freebsd/user/wa5qjh) joined #forth 03:44:50 --- quit: wa5qjh (Remote host closed the connection) 03:50:04 --- quit: KipIngram (Quit: WeeChat 1.4) 03:53:51 --- join: dave9 (~dave@90.20.215.218.dyn.iprimus.net.au) joined #forth 03:54:14 re 04:19:48 --- join: KipIngram (~kipingram@185.149.90.58) joined #forth 04:19:48 --- mode: ChanServ set +v KipIngram 04:24:58 hi KipIngram 04:46:47 --- join: siraben[m] (sirabenmat@gateway/shell/matrix.org/x-jwghojpcedciefsc) joined #forth 04:47:05 --- join: siraben (~user@unaffiliated/siraben) joined #forth 04:48:29 Hi Dave. 04:48:43 They rebooted my seedbox - had to get all connected up again. 04:49:01 I'm setting up a Matrix bridge right now, should be interesting. 04:49:06 test 04:57:27 --- quit: siraben (Quit: ERC (IRC client for Emacs 26.1)) 04:57:58 That looks promising, I guess. :-) 04:58:44 I'll have to get used to this graphical interface, it's a trade-off. 04:59:23 Yeah, I try to shun GUIs as much as possible. 04:59:31 The 16 character width limit on the calculator is ridiculous, I'm planning a re-write on the display system 04:59:32 So many circles 04:59:58 Um, isn't there just a limit on pixels that will put a cap on anything you do? 05:00:04 Or are you going to have it scroll? 05:00:11 16 is fine, for a calculator. 05:00:22 I guess. 05:00:29 There is a small font variant that is used in other programs 05:00:40 A few years ago I bought an "anniversary edition" calculator HP put out - HP-35S. 05:00:43 But then it becomes harder to read. I'll just manage space well. 05:00:53 --- join: nighty-- (~nighty@s229123.ppp.asahi-net.or.jp) joined #forth 05:00:55 It actually has the old-style key technology that my HP-41CV had back in college. 05:00:57 So nice... 05:00:59 --- join: dddddd (~dddddd@unaffiliated/dddddd) joined #forth 05:01:21 I haven't done a whole lot with it, but I got it out last night and fired it up. 05:01:38 My HP-42S emulator on my phone is far more capable, but the 35S is fun to play with. 05:01:59 I wonder what students will be using decades from now 05:02:07 There's also a "grass roots" calculator out there called the WP-34S. 05:02:33 It's based on some HP calculator and the community developed an alternate flash and a keyboard overlay and sticker set relabeling it. 05:02:45 It's generally considered the "most powerful non-graphing calculator." 05:02:54 There's pressure to not make calculators more powerful, because it could mean cheating. 05:02:56 I've got a couple of those too. 05:03:09 I find the graphic feature mostly a gimmick. 05:03:12 Well, they're already powerful enough for that. 05:03:21 Me too - I've never actually owned a graphing calculator. 05:03:31 My wife has an HP-48, though. 05:03:43 graphing* 05:03:44 I don't recall ever using it in tests. 05:03:58 I have sort of the same attitude there that I do re: GUIs. 05:04:05 I'm just a "text oriented" guy. 05:04:28 Technically it is possible to cheat using the approved calculators like the TI-84 05:04:45 Sure. 05:04:49 They ask you to clear RAM before tests, but there's nothing stopping you from archiving to flash and restoring 05:04:52 It's a computer, so. 05:05:08 I had a pretty open-minded professor once that said we could use our programmable calculators on the tests as long as we used only software that we wrote. 05:05:19 We had to let him inspect the calculator anytime he wanted to. 05:05:32 That makes sense 05:05:40 Back then there weren't that many upscale calculators, so he could plausibly know them all better than the students did. 05:05:41 Hah, I wonder if a Forth interpreter would pass. 05:05:45 That wouldn't be the case anymore. 05:05:52 I wrote a TON of stuff for that 41CV. 05:06:12 In what language? 05:06:18 What CPU did it use? 05:06:28 I implemented complex arithmetic, Gaussian elimination, and so on - I could actually just inspect an electric circuit, and do some complex math to set things up, hit a button, and have all the voltages and currents. 05:06:41 In its native programming language. Keystroke-based programming. 05:06:57 It didn't handle complex math antively - I implemented that. Had a "complex stack" in memory, etc. 05:06:57 Interesting. 05:07:04 The TI-84 supports that natively. 05:07:26 But it's bad at: factoring, displaying in other bases, arbitrary integer arithmetic, etc. 05:07:34 So I'd just visit each component in the circuit, do a bit of arithmetic around it, and hit a button to integrate that into the system of equations. 05:07:41 Once I'd done that for all components, I'd hit "solve." 05:07:54 It would crunch for a couple of minutes and spit back the phasor voltages and currents. 05:08:02 Coincidentally I'm learning about eletrical circuits right now 05:08:14 Cool stuff. :-) 05:08:20 RC circuits, rectifiers, transformers, etc. 05:08:22 Yeah, it is. 05:08:30 I also wrote a Smith chart program. 05:08:38 What's that? 05:08:42 Smith chart is a graphical tool used for dealing with transmission lines, waveguides, and so on. 05:08:45 --- quit: nighty-- (Quit: Disappears in a puff of smoke) 05:08:50 You "use it" in a graphical way. 05:08:59 Ah, so this is pretty useful stuff. 05:09:06 --- join: nighty-- (~nighty@s229123.ppp.asahi-net.or.jp) joined #forth 05:09:12 And I implemented it that way, so that there was an operational analog to all the "graphical things" you'd do with an actual Smith chart. 05:09:24 Another student once saw me using it, and asked about it. 05:09:32 When I told him what it did he was like "Can I have that?" 05:09:39 He sat there and wrote the whole program down. 05:10:18 I liked solving the problem that way, because (as opposed to just solving the equations) I didn't miss out on the "physical experience" of using the Smith chart. 05:10:32 So I still learned all the "what can you do with a Smith chart" stuff. 05:10:45 It depends on the problems being asked for as well. My calculator wouldn't be able to prove a group isomorphism, for instance. 05:11:11 I guess it was the same with the circuit solver - I was still going through the motions of setting up the equations and so on (node voltages, mesh currents, etc.) 05:11:19 The calculator just kept up with the details for me. 05:11:34 Sounds like a pretty advanced calculator 05:11:49 So those were "cooperative" tools - I still had to know what I was doing and understand the basics. 05:11:58 I just got rid of the tedious arithmetic stuff. 05:12:10 So I felt like that professor would have been fully satisfied with my use of the calculator. 05:12:21 It was the best there was in that day (circa 1982). 05:12:40 And of course there were the HP / TI wars we all waged on one another. 05:12:57 Sort of like vim and emacs - both were great, capable tools. 05:13:19 But I just committed myself to RPN and never looked back. 05:13:35 Now it's just TI 05:13:38 All the way 05:13:49 Yeah. HP just lost there way somewhere in the 90's. 05:13:52 Or late 80's. 05:13:55 But the thing is, a lot of people don't know how to use the more advanced features of the calculator 05:13:57 their 05:14:17 One thing I use the most is the built-in single variable equation solver. 05:14:21 I think in the hey day HP was still run by engineers - it basically still was the "garage workshop" company it had started out as. 05:14:31 But then the bean counters took over, and that was the end. 05:14:47 Type in something : "0 = ln(sin(x))-x^3" and it spits out the first answer for x 05:14:59 I use it all the time when I get an equation and don't want to manipulate it 05:15:09 Bean counters? 05:15:20 Yeah, sure. That's symbolic arithmetic, which is a step beyond mere numerical solving. 05:15:27 I never had a calculator that could do that. 05:15:34 But man, I could do the numerical stuff. 05:15:57 I wrote several different differential equation integrators for it as well. 05:15:59 Oh, it's not symbolic 05:16:01 Different core algorithms. 05:16:04 Just floating point 05:16:07 Oh, ok. 05:16:09 I believe it uses Newton's method. 05:16:14 Well, I wrote that for mine. 05:16:21 --- quit: smokeink2 (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 05:16:25 Did your calculator have floating point? 05:16:25 Stuff to factor polynomials, etc. 05:16:38 I don't know if it had hardware floating point. 05:16:43 But obviously it did floating point. 05:16:43 Ah, there's a built-in feature for polynomial factoring as well 05:16:49 Exclusively, in fact. 05:16:58 Same here, Z80 doesn't know floating point. 05:17:00 See, I could claim I understood that stuff, since I wrote the algorithms. 05:17:10 Right. Even integers are floating point until they hit > 2^33 05:17:23 In fact, knowing something well enough to code it tightly is a deeper level of knowing - you can't just half-know it and do that. 05:17:55 I never really needed to distinguish - integers were just special cases of floats for most of the work I did. 05:18:05 A lot of problems in my math/science exams require actual understanding of the topic, the calculator can only help crunch a bit. 05:18:41 Oh, and astronomy. I had programs to give me positions of planets at any time, convert right ascension / declination to hour angle / azimuth, and so on. 05:18:47 I had a real-time clock module. 05:18:54 So I didn't have to enter the time for that stuff. 05:18:58 Nowadays people use Python, I presume 05:19:10 It's getting to be that way, yes. 05:19:12 ^^^ a lot of what you said reminds me of the Emacs calculator 05:19:19 Which is a programmable, stack-based calculator. 05:19:25 I use it all the time. 05:19:31 One of the famous books is "Numerical Recipes." It started out presenting FORTRAN, but then later they issued a C version. 05:19:43 Wouldn't surprise me at all to learn tha there's now a Python version. 05:20:01 What does it include? 05:20:12 Well, I haven't looked at it in years. 05:20:40 Roots of polynomials, Newton's method, Gaussian elimination, eigenvalues, etc. 05:20:43 "The works." 05:21:05 Special functions, like Bessel's functions, spherical harmonics, etc. 05:21:14 Makes me appreciate the built-in features more. 05:21:19 And there's source code in there for all of it. 05:21:23 :-) 05:21:29 I appreciated learning to write them. 05:21:40 See yesterday's discussion of libraries. 05:21:41 See, I wish TI made their OS at least open-source, if not totally GPL'ing it 05:22:15 The Emacs Calculator manual is a fascinating read. It has all of what you mentioned. 05:22:22 You should go hunt down the WP-34S code. 05:22:25 And arbitrary precision fixnum 05:22:27 It may be open source. 05:22:38 Yeah, I want to write a bignum package at some point. 05:22:50 For Forth? 05:23:01 I'm thinking BCD-style arthmetic, once I understand the DAA instruction 05:23:23 My "proof of principle" problem on that will be to implement a Mandelbrot viewer. 05:23:31 I know it's been done, but I want to do it myself. 05:23:39 Don't do BCD. 05:23:44 Why not? 05:23:53 Use full integer precision, just arbitrary. 05:23:59 Like for me that would be N 64-bit cells. 05:24:05 It will be higher performance. 05:24:06 Oh, hm. 05:24:13 Same concept totaly. 05:24:17 I have 16 bit cells 05:24:28 So that's four hex digits per cell. 05:24:37 So it would be "binary coded hex." 05:24:42 But then you're not wasting any bits. 05:24:43 Yeah 05:24:50 But it's the same thing, in every way. 05:24:55 Just better. 05:24:57 0304 = 34 05:25:04 Right? 05:26:01 No, 34 would be 0034. 05:26:10 Well, you could do it the way you said. 05:26:12 Ah yes 05:26:14 But you're wasting half the bits. 05:26:23 And I'm just saying let that be *hex* 34, not decimal 34. 05:26:25 * siraben[m] uploaded an image: addr0.png (4KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/DhMHLCrGvVhFMdxPAZsoOxOG > 05:26:29 Apparently Matrix can send images, I'm checking to see what it looks like 05:26:49 Ah it appears as if I typed /me 05:26:54 Yeah, hex 34 05:27:29 $80, $AC, $46, $19, $18, $45, $80, $00, $00 = -4.61918458 Ã 10&44 05:27:37 10^44, not 10&44 05:27:48 Ok. 05:27:55 So from the third byte onward it's the significand 05:27:56 I didn't work it out, but seems reasonable. 05:28:12 No bits are wasted 05:28:40 I have a lot of the groundwork done for me, it's just a matter of reading user input into the floating pointer format. 05:28:41 If you use base 10 digits you waste bits. 05:28:53 Ah, base 10 woes. 05:29:01 I could just reimplement everything instead of relying on the calculator 05:30:13 Yes, you could. I likely would. :-) 05:30:25 That's just sort of how I roll. 05:30:57 I am using the available fpu instructions in my processor. 05:31:03 But that's relying on hardware. 05:31:21 My NUMBER will convert strings with a decimal point to 64-bit IEEE floating point. 05:31:27 Doing that "manually" would have been a nightmare. 05:31:28 It should be modular. i.e. I can rely on the OS for now and gradually phase to my own implementations. 05:31:34 I think I saw how to do it, but wow - involved. 05:31:38 But do I really need to redo floating point square root? 05:31:59 I just integer convert the mantissa and exponent and then calculate M*10^E using the fpu. 05:32:18 Ah, well no FPU here. 05:32:20 It's hard manually because the IEEE format has the exponent as a power of 2, not a power of 10. 05:32:36 I'll just stick with the OS then until I have more time 05:32:57 There's a lot of floating point ROM calls done for me, SIN COS TAN EXP SQRT LN LOG etc. 05:33:10 So if I were trying to implement floating point manually, for my own system, I'd probably cut a corner and store numbers as power-of-10 things. 05:33:23 Yeah, I'd be hesitant to give those up. 05:33:39 That wasn't quite what I was thinking of when I commented about "doing it yourself" a minute ago. 05:33:48 My end goal is this: eventually be able to make it through an entire test without leaving the interpreter at all 05:33:57 The calculator's internal format may not even be IEEE float. 05:34:07 "as if I were doing a test"* 05:34:11 Right, I don't think it is. 05:34:17 It's its own. 05:34:48 You can show that power-of-2 results in better preserved accuracy (less round off error) than power of 10, or power of anything else. 05:34:48 Fortunately, my math unit (set theory, group theory) doesn't need any calculator usage at all 05:35:08 That paper I linked you to a few weeks ago (what programmers should know about floating point) lays that argument out. 05:35:26 Yeah, I skimmed that paper but a more detailed reading will come 05:35:26 I need to learn more group theory. 05:35:32 I have only a vague understanding of it. 05:35:36 Haha, what for? 05:35:37 Sort of a "glimpse of the power." 05:35:50 It comes up in particle physics and other areas of advanced theoretical physics. 05:35:59 Well, a group consists of a set of elements and a binary operation that satifies four properties 05:36:08 associativity, identity, inverse and totality 05:36:09 Lie algebras and so on. 05:36:27 It's pretty intuitive, I think. 05:36:31 Yes, I have the basic concept down, at least. 05:36:42 Combines a lot of ideas you've had about various things 05:36:52 It may not even be group theory itself I need more of, but rather things it touches in its applications. 05:36:57 Like rotations and reflections, they form a group. Permutations are groups. Rubrik's cube moves are groups 05:37:07 Right. 05:37:14 Symmetry is a big deal in physics. 05:37:23 Are you a physicist? 05:38:19 No, I'm an engineer. PhD, though, and I crammed math in graduate school. 05:38:30 PhD in which area? 05:38:31 Specifically because I wanted to be able to self-educate in physics later on. 05:38:35 Ah I see 05:38:57 Engineering. My research was in electromagnetic theory - worked on electric guns and things like that (railguns, coilguns, etc.) 05:39:09 So engineering with a deep physics connection. 05:39:15 Wow 05:39:27 Did a lot of numerical modeling of em fields. 05:39:28 Maybe fields are more applicable to "real world" physics 05:39:32 Finite elements, etc. 05:39:49 As in mathematical notions of vector fields 05:39:57 Right. 05:40:00 Tensors are cool. 05:40:07 Physics just has a lot of math in it, it's hard to see what's most useful 05:40:10 Ah, I don't know what they are 05:40:14 You get up on top of tensor calculus and there's a "huge unification" of things. 05:40:20 Linear algebra is on my list of things to learn next 05:40:35 Also, Maxwell's equations seem like a lot of black magic to me 05:40:40 So many things that you learn earlier as special cases (like Coriolis forces and so on) turn out to be "just the plain phyiscs, properly expressed." 05:40:45 I love seeing special cases drop away. 05:41:12 When you get that, suddenly you see very plainly how gravity bends light ras even though they have no mass and so on. 05:41:43 You see how gravity affects EVERYTHING, because it's not really a force but rather is the shape of spacetime. 05:42:12 Linear algebra is VERY useful. 05:42:16 It's one of the cornerstones. 05:42:18 I'm learning a lot of seemingly unrelated things; Kirchoff's laws, induction, force acting on a moving charge through a field 05:42:28 All derived from maxwell's equations, right? 05:42:34 Right - Maxwell's Equations bring them all together. 05:42:45 Yeah, it's like finding out how calculus unifies a lot of kinematics 05:42:55 Kirchoff's laws are an "integral formulation" of certain aspects of Maxwell's equations. 05:42:59 Know of physics channels on IRC? 05:43:07 Not really. 05:43:32 In recent years my physics self-training has been primarily YouTube based. 05:43:50 Stanford has a great series of lectures, where Leonard Susskind teaches all manner of things. 05:44:15 I can't wait to have a long school break to finally get to the root of things 05:44:20 They're not just short "lay audience entertainment" pieces - they're full on college-course-length things. 05:44:39 And there's some absolutely golden stuff out there that Feynman did. 05:44:40 Ah, and not much prerequisite knowledge is needed? 05:44:43 I should watch more lectures 05:45:06 His 1964 or so Messenger lectures, and also a set he did in Austraila or New Zealand where he presented his theory, quantum electrodynamics, in a very "consumable" way. 05:45:29 I think the Messenger lectures were titled "The Character of Physical Law." 05:46:22 It's pretty fun watching how Feynman changed over the years. 05:46:37 In those 1960's lectures he wore a shirt and tie, was very clean-cut, and so on. 05:46:45 Much more casual in the Austrailan lectures. 05:46:56 Then I saw some he did much later, on computing, and he was almost a hippie. 05:47:16 But he was a master educator. 05:47:23 I'm guessing you were around to see the hippie movement? 05:47:39 Well, I was a kid. I was born in 1963. 05:47:47 So yes, but I didn't really "grok it." 05:48:09 My dad was a university professor, so I saw lots of college students in the late 60's, early 70's. 05:48:11 Oh wow, a lot of change since. 05:48:18 Oh yeah. 05:48:21 Interesting life. 05:48:48 My dad taught chemistry, particularly organic and bio. 05:49:12 So I had a pretty easy go of high school chemistry. 05:49:23 Haha, I'm having a pretty hard time with that. 05:49:27 I don't really remember much of it. 05:49:37 To the extent I do it's those parts of chemistry that are really physics. 05:49:41 Electron structure, etc. 05:50:18 The way quantum theory just tells you how the electrons will arrange themselves, and the way the periodic table just falls out of that, is pretty terrific. 05:52:10 It's kind of overwhelming seeing how deep all these areas of study are now. 05:53:19 I'd imagine it was much easier to get started with programming decades ago, because of the simpler nature of computer systems? 05:53:45 It's now non-trivial to set up a programming environment for a complete newbie 05:54:04 Sometimes it's like a Catch-22, to know how to program, you need to know how to program 05:54:17 I think so. When I was in college (at least in EE) we learned a lot about processor architecture, and so on, and the idea that it's just this state machine interpreting a set of bit patterns in consecutive memory locations was pretty straightforward. 05:54:29 There were no cache memories, there wasn't really much in the way of multiple cores, etc. 05:54:36 So it was a pretty clean, clear setof ideas. 05:54:38 many programming newbies take a lot of the "stack" for granted and never look below their high-level programming language 05:55:44 I remember the first time I heard the term "stack" and "heap" was in dealing with C 05:56:04 But even C has a lot of confusing things for beginners 05:56:20 Undefined behavior, little vs. big endian, byte alignment 05:56:22 (in this case i meant the layering of different parts of technology) 05:56:34 Ah, that too 05:57:02 Are there still microcomputers on the market? 05:58:14 if you mean keyboard-integrated computers like the C64, arguably that's what laptops with a broken or missing screen are ;) 05:59:44 Well, I don't see those things as being "C" things. 05:59:51 The compiler will handle those things. 06:00:05 I guess if you're trying to cast strings to ints, you need to know that stuff. 06:00:12 But you can learn a lot of C without ever doing any casting. 06:00:44 As long as you're just using types that you declare the compiler will handle that stuff. 06:04:37 I think there's great value in understanding a certain amount about your underlying hardware, though. 06:05:09 For example, knowing how your cache works. There are some array processing algorithms that will run nice and fast if you process by rows, but suck if you process by columns, because you don't get the same kind of cache benefit. 06:05:21 Well, there's that 06:05:24 So knowing how the array is stored and how the cache works lets you get that right deliberately. 06:05:47 And branch prediction 06:06:00 I think there's too much of a perspective that softwareruns in a "virtual environment" without any real connection to the "underneath." 06:06:05 Yes, that too. 06:06:08 For sure. 06:06:31 Back when I started, you could look up the cycle count for each instruction in a book and add 'em all up, and that was that - that's how long the program too to run. 06:06:39 Like it was a clock. 06:06:43 Same here on my calculator 06:06:47 But not anymore 06:06:59 Totally not the case today. 06:07:00 Right. 06:07:09 A big problem I'm having is what is "data" and what is "code"? 06:07:22 I'm writing a disassembler, and it happily "disassembles" strings and so on 06:07:29 The underlying hardware is just as precise as ever - it's just that the events that can occur are so much more involved now. 06:07:33 code is what you execute, data is what you dont, there is no clearer definition 06:07:57 ah, a disassembler needs heuristics or user input :| 06:08:05 Ah, I see 06:08:09 Technically SEE is a disassembler. 06:08:20 But I'm taking about disassembling binary 06:08:21 Yeah a dissassembler needs to be given a valid starting point. 06:08:28 So I can see what the ROM calls are all about 06:08:38 And cant always recognize when it stops seeing code and starts seeing "not code." 06:09:21 Well, when an unconditional RET is reached maybe 06:09:22 Or JP 06:09:24 Or really, I guess, can't EVER recognize that unless you've told it a way that will work in your system. 06:09:35 Possibly, if there's just one exit point. 06:10:01 If the dissassembler was smart enough, it could see conditional jumps and know that where they went was more code. 06:10:13 But some kind of intelligence has to guide that process. 06:10:23 To some extent you can code that in, but sometimes not. 06:10:59 Even if your dissassembler was very smart (followed conditional jumps, etc.), you could probably trick it. 06:11:15 I imagine I could pretty quickly write something that had a conditional jump that was never taken in operation. 06:11:22 And just have it point into the never never. 06:11:30 Dissassembler would lose its mind. 06:12:12 never never == Dresden Files reference. 06:12:22 Best books EVER. 06:12:33 that's the idea behind the "interactive disassembler" (https://hex-rays.com/products/ida/index.shtml for example) 06:13:04 the user can think a little better than a set of heuristics, so it can correct the heuristics' mistakes 06:13:06 I think there are a lot of problems that are best solved via a "collaboration" between human and computer. 06:13:22 yep 06:13:54 I'm pretty against a lot of machine learning stuff, because it's inherently a "black box" 06:14:03 I prefer correct programs by construction 06:14:19 Yes, and such black boxes make some really profound errors at times. 06:14:24 Not a linear algebra monster tuned by throwing data at it, although it might be "good enough" 06:14:32 There's a famous case where a learning algorithm was trained to discriminate dogs and wolves. 06:14:43 It misidentifed a dog as a wolf, and the engineers wanted to know why. 06:14:43 "heuristics" doesn't imply machine learning 06:14:54 So they put in code to "white box" the algorithm - show what parts of the picture drove the decision. 06:15:03 Turned out in that case it was snow in the bottom corners of the screen. 06:15:04 Also Google's Photo AI confused black people for gorillas, because of a bias in the data set 06:15:12 They had to remove "Gorilla" as a category 06:15:14 Without realizing it, they had correlated woves with snow in their input pictures. 06:15:26 Have you read GÃ¶del Escher Bach by any chance? 06:15:28 Best book yet 06:15:38 Yes, 15-20 years ago. 06:15:44 That's a LONG book. 06:15:57 Took a couple of months of on-and off 06:16:02 But I totally buy into the idea that there are things a human mind can "just see" the truth of that a machine can't prove true. 06:16:05 A bit heavy on mathematical logic, for good reason. 06:16:09 Yeah, LONG but worth it 06:16:25 I see GEB as just the quantitative "backup" of that idea. 06:16:36 I particularly like the demonstration of the A.I. program "ETAION SHRDLU", something like that 06:16:44 Or PLANNER 06:16:53 I don't remember it to that level of detail. 06:17:11 It was a dialogue between a human and program 06:17:18 "Move the blue cube on top of the red cylinder" 06:17:20 etc. 06:17:28 Ah, Ok. 06:17:52 Machine learning won't solve every problem, but it's a good start 06:18:02 Something deeper is missing in our notion of intelligence. 06:18:10 I think we'll continue to get better at making machines "mimic" human thinking processes, but I think it will always be a "mimic" and never the real thing. 06:18:17 Unless we start making computers that are more like brains. 06:18:28 Organic, maybe, or possibly quantum - don't know exactly. 06:18:30 Perhaps the answer is to simluate a brain? 06:18:36 But clearly humans can construct thinking machines. 06:18:38 Yeah, a hybrid might just work 06:18:47 It hapepns all the time, and the initiation of the process is FUN. 06:18:52 A chip on a brain, or vice versa 06:18:52 They're called babies. 06:18:58 Haha 06:19:20 Yeah, A.I. is just a naturally exciting field. 06:19:22 So I won't claim we'll never figure out how to do that using some other manufacturing technology. 06:19:25 But severly overhyped nowadays 06:19:33 But I feel pretty sure it won't be standard computing as we know it today. 06:19:40 "Look! A phone with an A.I. chip that's actually a GPU in disguise!" 06:20:12 I think these days a better name for "AI" would be "advanced probabilistic algorithms." 06:20:20 It's neat stuff. 06:20:23 But it's not "thinking." 06:20:30 Not aware. 06:21:08 And you've got really scary applications - like some state up in the northern part of the US has an AI program they use to recommend sentences in criminal cases. 06:21:12