00:00:00 --- log: started forth/05.10.03 05:37:39 --- join: virl (n=hmpf@chello062178085149.1.12.vie.surfer.at) joined #forth 05:55:30 --- quit: LOOP-HOG (Read error: 110 (Connection timed out)) 05:58:34 --- join: Ray_work (n=vircuser@adsl-65-68-201-174.dsl.rcsntx.swbell.net) joined #forth 05:58:42 --- nick: Raystm2 -> nanstm 06:09:21 --- join: madwork (n=foo@derby.metrics.com) joined #forth 06:30:50 --- join: PoppaVic (n=pete@0-1pool73-110.nas24.chicago4.il.us.da.qwest.net) joined #forth 06:31:08 Howdy 06:50:53 hi PoppaVic! :) 06:51:01 Howdy-doody 06:51:03 ready for a productive day 06:51:08 Sure tryin' 06:51:10 ? 06:51:13 hehe 06:52:24 I've circled back to considering either a tag-field/stack-entry or a hidden-array of byte; at least for the control-stack. 06:52:28 * Ray_work finishes ( for the most part ) all of his cf projects. Time to document them, then put them to rest and move on to a 'real' language. 06:52:51 'cf'? 06:52:57 colorforth 06:53:00 ahh 06:53:18 I spent a year developing cf apps, time to give up the ghost there. 06:53:52 my robot/newbie-forth-trainor complete, just the docs to go. 06:53:54 It has occured to me that using tags (somehow), for at least the flow-stack, would simplify flow-control code 06:54:57 crc has some pretty neat flow-control structures in his retroForth. 06:55:02 I'm not sure exactly what tags we need, but it seems likely that it's a very small set. 06:55:21 I'm certain you are correct. 06:55:57 at the minimum, markers for 'returnptr' and 'number' may well be all that is needed 06:56:39 or, even more generically: 'code' vs 'number' 06:57:12 I see. 06:57:50 This would, I believe, greatly simplify the idea of 'exit', 'leave', 'unloop', etc - we'd not NEED 'unloop', if we use a decent set of tags 07:03:07 Back shortly... 07:04:15 okay 07:21:21 back 07:39:14 The longer I stare at do-loops the more they piss me off 07:40:43 in fact, the info-file suggests about 5 different reasons to wish for a bat and a quiet alley. 07:42:08 I'm beginning to see a similar issue with the idea of "it's just a CELL' when we talk about stack-entries. 07:45:40 Crap. I hate to admit it, but we got warts all over the place. 07:47:17 --- join: sproingie (i=foobar@64-121-2-59.c3-0.sfrn-ubr8.sfrn.ca.cable.rcn.com) joined #forth 07:47:57 Assuming merely a CELL-width stack... You've got: Signed, Unsigned and even 'intended-width' (range). 07:48:08 hi sp 07:48:24 mornin 07:49:56 sproingie: I presume your own code has some issues with 'do-loops' as well? Gforth added at least 3 more I never thought of (and should have) 07:51:54 i'm not sure what gforth's looping constructs are 07:52:10 i'm working on getting THROW and CATCH working in rf 07:52:47 well, they mention that "DO" is pretty much deprecated in favor of ?DO - and then also mention THAT has "issues" so they added +DO and U+DO, etc, etc - it's quite a mess. 07:52:52 seems to be doing the flow transfer to the handler correctly, but it isn't cleaning up the stacks 07:53:17 A lot like a goto, then? 07:53:18 THAT? 07:53:21 never heard of THAT 07:53:26 ohhh 07:53:32 that as in ?DO 07:53:36 right 07:53:37 i thought since you capitalized it ... ;) 07:53:41 sorry 07:54:04 throw/catch is more like a nonlocal return 07:54:21 flipside of goto/longjmp 07:54:30 it's exactly like setjmp/longjmp 07:54:53 yeah, the info on setjmp/longjmp never excited me. 07:55:17 I've used it a couple times in my life - but only by screaming and just throwing it at the code. 07:55:25 they're rather limited, and usually need a lot of glue to keep stuff from crashing 07:55:40 solaris has getcontext/setcontext, which aside from having better names, are also more well behaved 07:55:50 i believe linux appropriated those also 07:55:51 Their docs on what it sets up and does is pretty damning 07:56:04 I've heard of those... somewhere 07:56:25 they save enough info to use them even as downward continuations. or so i'm told 07:56:49 somehow i don't see that, since they'd have to save a big chunk of stack 07:56:50 I can't think in "continuations" 07:57:11 yeah, there is an awful lot of stack and even register foo-foo therein 07:57:39 well for upward continuations like catch/throw, you don't need to mess with the stack 07:57:54 if you were counting on registers to be saved tho, forget it 07:58:21 setjmp/longjmp make pretty lousy multitasking primitives in that regard 07:59:04 i wonder what the instruction set of the cell is like 07:59:12 what "cell"? 07:59:19 would be nice to get forth running on one of those. maybe one on each of its units 07:59:29 ahh a cpu 07:59:50 the cell, new cpu by ibm everyone's been drooling over, playstation 3 is using it 08:00:18 Well, I'm just about convinced I can all but discard double-number codes at this level, and that I need to add a CPU-emulator level. 08:00:21 looks very SIMD or vector pipeline based, possibly both 08:00:49 has one main CPU, and 7 APU's (auxilliary processing units) 08:01:13 so, it's a star/net/hex-set 08:01:23 or maybe 8 apu's ... i know sony is disabling one of 'em for higher yields 08:01:34 THe idea has been around for years. More power to IBM 08:02:07 oh yeah it's not as new or strange as a VLIW like the Itanic 08:02:19 though i imagine it consumes about 1/1000th as much power as one of those 08:02:37 One of the issues always was that you'd need a special asm/compiler which had rules internalized to opt-off some load 08:02:52 which makes it a pretty decent fit for the PS3 08:03:16 since the PS2 already worked like that. it took developers years to get used to it but theydid 08:03:21 I have to ignore special CPU's - I can't afford to even dream about such things 08:03:36 PS2 has one CPU, two VPU's and the GS 08:03:41 and they're all more or less independent 08:03:48 no, I meant the Cell 08:04:03 (and I don't even own any games) 08:04:37 the model that dev kits use for the PS2 will probably work for the PS3 pretty well 08:05:09 hopefully sony will document the FCS release in English this time 08:06:11 oh, hey - I was going to ask you.. 08:06:39 That zero-packing-terminator? Is that even used for variables, constants and such? 08:07:05 I know it's prolly more jason's issues, but I thought I'd ask you 08:07:58 Seems unlikely, but I got a bit confuzzled out of the gate this AM 08:08:37 zero-packing terminator? not sure what that is 08:08:55 is that like telling the compiler not to pack a struct? 08:09:20 yeah, it must be specific to tathi/jasons code. They use it to tell the code-fetch that no more opcodes follow. 08:09:56 you can ask a compiler to not pad structs at the end to an 8 byte boundary, but i don't think you can keep it from aligning on a 4-byte boundary 08:10:04 they fetch a CELL at a whack, and 1+ can be run, but they hit the zero and end there 08:10:40 sproingie: yeah, sorry - not C/asm packing - I'm pretty sure I don't much care for the idea. 08:10:53 i wouldn't rely on a compiler respecting a "not packed" directive 08:11:17 I don't like using gcc "metacompiler" directives at all 08:11:33 most compilers have something similar 08:11:43 sure, and I'd still avoid them 08:12:07 probably a good idea 08:12:14 The only time I can envision getting that involved would be like for tathi's object-file header-stuff 08:15:20 has anybody here swiftforth? or a trial version? I don't like that register stuff, so that would be really nice if someone could give me a little version for testing. 08:17:11 I don't know why the hell every software company wants a formular with my data when I'd like to test their software. I think it's idiotic. 08:18:42 ok, how is swift forth? 08:23:39 quartus lets people download without getting nosey 08:24:15 it's your basic shareware. registration just to download always struck me as self-defeating 08:24:35 like marketing wants all its info up front so they turn people off right away 08:28:23 * PoppaVic screams, sighs, "ah-hahs".. And then ponders How_To... 08:29:30 OK, yeah... Bastards. *sigh* I will need to create an opasm table to flog and access all the pseudo-regs. 08:30:17 I think I solved my problem with different cell sizes in xell. 08:30:47 Given a decent set, THIS would be the "layer" that folks would optimize for a specific CPU. 08:31:18 --- quit: Quartus (Remote closed the connection) 08:40:12 --- quit: warpzero ("leaving") 08:40:19 --- join: warpzero (n=warpzero@wza.us) joined #forth 08:51:47 --- quit: warpzero ("leaving") 08:52:20 --- join: warpzero (n=warpzero@wza.us) joined #forth 08:52:21 PoppaVic: it's just a terminator, rather than requiring the VM to keep a count of where it is. 08:52:36 ahh, ok. 08:52:55 so, every bytecode-seq would need the \0? 08:53:18 you want to fetch several opcodes at once (because of alignment issues - you can only a literal from certain addresses) 08:53:33 right 08:53:53 no, that's why I chose zero - I just shift the instruction word down, and the rest gets filled with zeros. 08:54:07 so when you hit the end...you know you need to fetch another instruction word. 08:54:09 ohh 08:54:21 and embedded data? 08:55:11 The only kind of embedded data that my VM knows about is single-cell literals. 08:55:30 hmm...this needs a diagram :) 08:57:02 Yeah. versus the fetcher/decoder 09:07:06 * PoppaVic rotfls as he works on the pseudoasm-layer 09:10:29 tathi: to be honest, it sounds like a small caching mechanism 09:11:16 And, thus.. I have to wonder if the shifts cost more than just indexing. 09:15:48 yes, but what alternative do you have? 09:15:54 where do you put literals? 09:16:01 right 09:16:20 I know what yer saying, I've been glaring at it for a week 09:16:47 actually, we had a major oversight when we did Fovium -- offsets for branches/calls should be IN the instruction word. 09:16:58 I don't think Jason has gotten around to switching it yet though. 09:17:01 ..but, at some point 3+ access are going to be unaligned or shifted 09:17:20 Yes. I am glaring at that right now. 09:18:23 We already know that any silicon is faster, but grossly variable. So, I am oddly leaning toward emulating a forthish CPU again. 09:18:43 this probably isn't any help, but I snipped out just the "execute loop" for Fovium. 09:18:46 http://rafb.net/paste/results/3W5Tvf84.html 09:19:20 And, because of unixish/posixish systems, there is really no point in considering interpretive-compiling/execution of the "asm" 09:20:51 I should say "real asm" 09:21:13 tathi: yeah, sorta' akin to my loeliger book 09:22:49 tathi: if it matters much, cogitate this some and email me... I think we need to consider a pseudo-CPU and what commands we need for it. If we can emulate an 'ideal' CPU, the rest will fall into place 09:23:45 I have been. I'm coming to the conclusion that there is no such thing as an 'ideal' CPU, so that it makes more sense to try and build a flexible compiler infrastructure. 09:24:17 I'm beginning to suspect we need [in]direct, {src,dest}regs, and some basics 09:25:27 tathi: well, we can't be harmed then.. All we care is our bytecode runs all over, and the interface to true asm/C or whatever does what we want to the pseudo-cpu 09:26:20 All we care is that the bytecodes/lists/structs hold up. 09:27:05 and, hell: maybe we will end up writing >native compilers per-platform 09:27:06 I just have no interest in something that runs all over (yet?). 09:27:22 I'd rather put the effort into a decent compiler for my box, and then generalize as I need it. 09:27:33 well, sure - I can live in C for the same reasons 09:27:43 ..now, reconfigure it all the time 09:28:31 I'd rather "compile" bytecode and leave optimizing nativecode compilers as an aside 09:29:58 I have to agree with a buddy of mine from Long Ago - "Forth is the worlds most unsupported Language" - it's worse than pulling teeth to get folks to remotely agree. 09:31:08 I haven't switched it yet 09:31:10 Actually, it's a lot like sitting in #asm and trying to folks 09:31:22 to [talk with] 09:31:26 I might only switch it for calls 09:31:42 so branches will be easy to patch 09:32:52 it's not that much harder to patch branches if they're in the instruction word... 09:32:54 but whichever. 09:33:18 yeah 09:34:06 >>r @ -64 & swap 6 << or r> ! 09:34:20 heh... I begin to see why folks charge for their "portable" Forths 09:34:41 vs: ! 09:35:06 vs everyones cpu-specific whores ;-) 09:36:16 oh, what am I thinkig 09:36:25 ah, well *sigh* - it's par for the course 09:36:26 don't have to grab the bits, just patch in the call opcode 09:36:45 It seems to me that with a compiler, you could easily add more "primitives" 09:36:49 here 6 << op-call or swap ! 09:37:00 tathi: yep 09:37:02 Whereas with a bytecode engine, you really want to get everything designed properly before you build it. 09:37:12 assuredly 09:37:20 And...seeing as how I'm not a fan of big-design-up-front... :) 09:37:34 tathi: the kicker remains that objectfile versus pcodefile issue. 09:37:41 tathi: I don't have a problem changing the vm as I go 09:37:58 tathi: especially adding and removing opcodes 09:38:01 JasonWoof: oh, I meant as an engine for other people to build stuff on top of. 09:38:11 tathi: in forth it doesn't seem to matter much what's an opcode, and what's a word 09:38:18 true. 09:38:21 tathi: ahh. I've not read the backlog 09:38:37 Yep. We need a version of forth that is more like CINT and C-- 09:38:47 yeah, whatever API you give people better not change much 09:38:55 exactly 09:38:55 That's what I'm aiming for. Moving at a glacial pace though... 09:39:05 lunch, bbiab 09:39:11 sure 09:39:13 ooh :) lunch 09:39:16 sounds good. 09:39:24 * JasonWoof puts on some clothes and heads for the kitchen 09:48:37 brb 09:48:48 --- quit: PoppaVic ("makes a few quick calls and returns") 09:57:33 --- join: aardvarx (n=folajimi@shell2.sea5.speakeasy.net) joined #forth 09:57:43 --- join: PoppaVic (n=pete@0-2pool198-117.nas30.chicago4.il.us.da.qwest.net) joined #forth 09:57:48 greetings, all. 09:57:48 heh. looks like I started something there :) 09:57:52 hi aardvarx 09:58:13 hi tathi 09:58:14 OK, back... I think we need a major glare, yeah 09:59:04 lol. I hadn't seen CINT before. 09:59:13 that's amusing. 09:59:28 tathi: I know you've generated asm to create direct objects. Yeah, CINT is one of several. 10:00:06 Note that I've only generated executables. Never quite got around to relocatable (.o) files. 10:00:16 ahhhh 10:00:49 It doesn't look too terribly difficult, but I've never actually done it. 10:00:51 Well, I look forward to seeing object-file code. 10:01:34 Seems almost certain that we'd benefit from a synthetic-assembly in forthish and interp+generator/compiler. 10:01:38 Yeah, I will get to it eventually. 10:02:54 I'd like to solicit comments on registers/use/opcodes, though.. I'm gonna' add a layer below what I have to talk strictly with the VM and its "registers? 10:02:59 ?/" 10:03:48 the "asm opcodes" need intense folding, which is why I called it a day early. 10:04:51 direct|indirect, sourcereg/address/destreg, etc. 10:06:49 The kicker with C remains: actual object-code is closed - fine. We want an emulator that we can forth-compile/interpret for. A second program to generate local-asm and objects/libs/modules is fine with me. 10:07:45 and, we may well add some "pseudoasm" to support dlopen-type shit. 10:08:21 we don;t NEED to fully parallel a cpu, we only need to parallel enough to interface most places. 10:09:11 My biggie is: we don't reflect the platform-cpu, just our own "system" 10:12:55 I think the asm-opcodes should be folded, spindled and mutilated as tight as we can (like the 4-byte fetches emulate) 10:13:39 and, we then have - almost universally - merely one set of opcodes/level/table that really demands "porting" 10:14:24 It mostly takes fig-forth DOWN one-level 10:15:18 And, mind you: I don't WANT to accept inline-asm (true asm). 10:15:42 Let folks use linkers for that. 10:17:03 It's just occured to me, too: this may simplify the RStack. 10:19:39 I need more feedback on "what 'asm' set we need", but it seems likely we can generate "slower" bcode (and who cares, nowadays?) - but simplify actual compilers no end. 10:20:59 I'd even appreciate some code-kibitzing on that "asm codeset" It's not like brain-surgery. 10:21:47 anyway, the hell with it for today... I'm gonna' chow and snooze. 10:21:57 talk to you tomorrow 10:21:59 --- quit: PoppaVic ("Pulls the pin...") 10:42:10 --- join: Snoopy42 (i=snoopy_1@dslb-084-058-132-076.pools.arcor-ip.net) joined #forth 11:29:47 --- join: zoly (n=l@tor/session/x-dd7e61ce1d0e3684) joined #forth 11:29:56 'morning 11:33:43 Hi, zoly. 11:33:55 What are these tor/session/* masks? 11:34:54 tor.eff.org ? 11:35:50 right 11:36:04 Thanks. 11:38:03 prevents irc scanners to trace back to the system one connects from 11:38:03 i found the number of portscans increasing when i connect to irc nets 11:39:38 Looks pretty nice. Is it efficient? 11:40:08 you wouldn't want to use it for web browsing all the time 11:40:32 but with a proxy switch on your web browser, it is nice 11:41:42 the entrance behaves like a socks4/5 proxy 11:42:52 for irc, it has a bit the function of cloaked accounts, without the need to hassle freenode staff 11:43:05 and works on other irc nets too 12:20:56 --- join: Quartus (n=trailer@ansuz.pair.com) joined #forth 12:24:33 Quartus!!! 12:24:35 Howdy! 12:26:21 Hi. 12:26:32 Hi there. 12:27:09 How's the business doing? 12:27:41 Always room for improvement. :) How are things with you? 12:32:01 Not too bad, not too bad... 12:32:34 Can't seem to get the motivation to code :( 12:32:47 Not a problem I usually have. 12:33:11 Oh? 12:34:38 Yes, I enjoy the process, so I don't have to reach deep for motivation. Is there something specific you're wanting to write? 12:34:44 money counts. 12:34:56 I'm still learning C. 12:35:05 It helps to have a project in mind. 12:35:06 So, I am reading the book I have. 12:35:14 It's not a bad read... 12:35:31 but it does not make me giddy, either. 12:35:39 Maybe you need a better book :) 12:35:42 It also helps to have ONE project in mind. So you don't spend the day trying to pick a project. 12:35:54 I prefer to have several going at once. 12:35:54 Robert, I'm not that good yet. 12:36:19 Are you learning C so that you can do a specific thing or things, or just because you think it'd be a good thing to learn, in abstract terms? 12:36:23 I mean, "hello world" is not a problem, 12:36:30 Quartus, the latter. 12:36:47 aardvarx, that makes it hard. It's a means to an end, not an end in itself. 12:37:08 What can I do? 12:37:24 As I said, having something specific you want to achieve would help a lot. 12:37:56 I've got the "Hello world" thing down... but I don't know how to "use/write sockets" 12:38:05 * aardvarx wonders what a "socket" is... 12:38:05 Do you need to use/write sockets at this point? 12:38:12 Nope. 12:38:19 Then that's of little consequence, at this stage. :) 12:38:35 Here's a thought: 12:38:57 This might sound absurd, but please hear me out 12:40:23 I noticed that some buildings on the college campus I attended had Greek symbols on them (I don't know what is means, but they are usually over the door of the building or something...) 12:40:52 Anyway, I was wondering how many combinations of those groups can exist. 12:41:41 I was thinking I would write something that will spit out a combination (I noticed they usually are two or three letters per building...) 12:42:08 That's a pretty simple calculation -- number_of_letters_in_the_greek_alphabet**3 12:42:23 Also, there should be a way to keep track of combinations that are already used. 12:42:40 So, what do you think? 12:43:05 Do you want to generate an exhaustive list of all two- and three-symbol strings in the Greek alphabet? 12:43:36 Yes. Perhaps even include single-characters as well. 12:44:05 How much experience do you think that requires? 12:44:23 Also, is it something that can be implemented entirely in C? 12:45:12 The whole thing about "keeping track" suggests that I might need to use a database. Perhaps that might mean I need to learn MySQL? 12:45:39 aardvarx, you can write a *very* short program that prints all of them in order. 12:45:55 Short? 12:46:00 :) 12:46:01 In pretty much any language, though if you want them displayed properly you'll have to sort out fonts, etc. 12:46:22 Yes. Do you think it would be a long program to print out all numbers between, say, 000 and 999? 12:46:44 lol.. 12:46:54 I think at this point, I'd settle for just spewing out things like "Alpha", "Beta", "Gamma", etc... 12:46:57 * zoly thinks about installing oracle for managing the telefone numbers of his contacts 12:47:12 aardvarx -- then that takes away the worries about the fonts. 12:47:16 You know that many people? 12:47:23 * Robert has a ~/stalk.txt file. 12:47:24 at least 50 :) 12:47:24 zoly, I get it. The My 12:47:39 "MySQL idea was ridiculous" 12:48:37 There are (I believe) 24 letters in the Greek alphabet. So there are 24*24*24 possible 3-letter strings -- 13824. 12:48:49 576 two-letter strings. 24 one-letter strings. 12:49:05 In total14424. 12:49:14 * aardvarx queries google for Greek alphabet... 12:49:17 is there an idiom like "shooting sparrows with cannons" in english? 12:49:29 zoly, killing a fly with a sledgehammer 12:49:50 sound like a good one 12:50:06 Quartus, it is twenty-four. 12:51:04 yup. sci-fi authors tend to futurize that to "swatting a fly with a laser cannon" :) 12:51:13 Quartus, so here's an idea: First, I start out with your suggestion of just "spewing everything out." 12:51:17 It is an extremely simple program to print an exhaustive list of all one, two, and three-symbol Greek strings. No database is required, any more than you'd need one to print the numbers from 0 to 14424 without doing one twice. 12:51:46 How about subtracting those that are already used? 12:51:56 If you're printing all of them, no subtracting is required. 12:52:20 actually, the "shooting mosquitos with laser" idiom was used by C.M. after the NC4000 came out. personal SDI 12:52:46 :) 12:53:13 Quartus, I guess that would be the next step would be to find out which string has been "taken" already. 12:53:18 though there was speculation about whether he was commenting on military applications of NC4000 12:53:41 No need to do that either, aardvarx. A three-symbol Greek string is no different than a three-digit number, except it's in base 25 instead of base 10. 12:53:44 "I guess that the next step would be to find out which string has been "taken" already." 12:54:10 Base 25 ??? 12:54:20 Wait, base 24? Sorry. Base 24. 12:54:28 aardvarx: first simplify problem, then start coding :) 12:55:00 zoly, I want to make sure that I have the tools/qualifications. 12:55:12 aardvarx, yes. Decimal digits are from 0-9; ten different symbols, hence 'base 10'. Greek letters are from alpha to omega; 24 different symbols, hence 'base 24'. 12:55:24 aardvarx: you think too complex 12:55:45 zoly, how did you know??? 12:55:51 :P 12:56:00 aardvarx: you told us 12:56:07 I did? 12:57:30 Quartus, how does the issue of "Base" factor into this conversation? Could you please explain? 12:57:48 Did you understand the last thing I wrote? 12:58:07 No :( 12:58:39 Ok. Assign alpha a value of 0. Assign each successive letter the next highest integer. 12:58:45 What is omega's value? 12:58:49 23 12:59:11 Ok. So if I said your greek string is represented by {0,23,0} -- what's the string? 12:59:32 Alpha, Omega, Alpha ! 12:59:35 I see! 12:59:48 Hmmm. 12:59:56 Ok. Now working in decimal, if I said your decimal string is represented by {5,4,2} -- what's the string? 13:00:58 Epsilon, Delta, Beta ? 13:01:17 I think you're off by one there. 13:01:27 Decimal. Forget Greek. Assign a value of 0 to the symbol '0', and so forth. What is the value of '9'? 13:01:59 aardvarx, the conceptual piece you may be missing is that decimal numbers are written in an alphabet of ten symbols, '0' '1' '2' ... '9'. 13:02:21 Yes! 13:02:29 Zeta, Epsilon, Gamma ? 13:02:45 A number, written out, say -- '542' is nothing more than a three-letter string in the 'decimal' alphabet. 13:03:37 Hold it! 13:03:39 A three-letter greek string, say 'Alpha Omega Alpha' is nothing more than a three-digit number in the 'Greek' alphabet, where the digits aren't named '0' ... '9', but 'alpha' to 'omega', and there's 24 of them. 13:04:02 That's not the same as {5,4,2} though. 13:04:05 Is it? 13:04:12 {5,4,2} in decimal is '542'. 13:04:42 {5,4,2} in Greek is 'Zeta Epsilon Gamma'. 13:05:03 Quartus, I guess where I run into trouble is {20,15,22} 13:05:14 What kind of trouble? 13:05:20 201522 is not a valid entry. 13:05:39 It is greater than 24*24*24 13:05:39 Not a valid entry in what context? 13:05:47 I'm going to try to simulate a fairly large DTL (diode-transistor logic) digital circuit, mainly interested in propagation delays and their effects. What I'm wondering about is how the number of inputs in an NAND gate affects the output propagation time. Anyone experienced with this? 13:06:02 aardvarx, it's 20*24*24 + 15*24 + 22. 13:06:13 ok, in decimal, you have 10 'letters', so each 'digit' is worth 10 times more than the next one. 13:06:29 I see. 13:06:31 in Greek, you have 24, so each digit is worth 24 times more than the previous one. 13:07:12 Aha! 13:07:33 So, if I get 10245 as an input... 13:08:23 zoly is right. 13:08:28 * aardvarx is thinking too hard. 13:08:35 I lost my train of thought. 13:08:44 I was going to say divide by 24*24... 13:09:22 That doesn't really say anything about the first letter. 13:09:48 tathi, wait a minute... 13:10:06 Are you saying that it can be reduced to a basic arithmetic problem? 13:10:18 aardvarx, yes. You can enumerate any string. 13:10:30 Aha! 13:10:38 aardvarx: I was just saying in words what Quartus put in an equation. :) 13:10:39 So you can iterate over the entire integer space, and generate each string in turn. 13:11:07 I was thinking that I would have to write all the possible combinations out by hand! 13:11:30 No. 13:11:49 No more than you'd have to write out each three-digit decimal number by hand if I asked you to print all the numbers from 000 to 999. 13:12:34 So do you see how you wouldn't need to keep track of those already printed? 13:13:39 Only if I don't care to keep a record, yes. If I am just counting... 13:13:54 all I have to do is keep moving forward. 13:15:27 Right. So SQL -- overkill. :) 13:15:47 Hmmm. 13:18:37 So, if I am given 12345, I will divide it by 24*24; then I will reference the remainder to the alphabet. The returned value would be the first letter (a.k.a Most signficant bit) in the collection, right? 13:18:45 even "keeping a record" would be simple 13:19:42 zoly, I won't argue with you (because you probably know much more than I do at this point,) even though I don't believe you. 13:20:26 you know already how to enumerate any string. i.e. representing each of them with a unique number 13:20:27 zoly, I guess *simple* is relative :P 13:20:53 use this number as an index into an array, where you mark "used" or "unused" 13:21:58 zoly, where does the information in the array come from? 13:22:03 aardvarx, you'd take 12345 mod 24 (in C this is %24). That's the lowest digit. 13:22:24 Then you'd divide 12345 by 24, discarding any fractional component. 13:22:33 aardvarx: just true vs false, or 1 vs 0 13:22:33 Then you'd take the result % 24 again, that's your next digit. 13:22:39 Repeat as requried. 13:22:42 *required 13:22:57 Goodness gracious!!! 13:23:05 you start with "all false" = non used 13:23:15 Why on earth did I start from the other end? 13:23:18 none 13:23:44 zoly, I see your point, but I.... 13:23:50 Aaaah! 13:23:50 aardvarx, note that you can extend this technique to print a value in any base. Binary = base 2, decimal = base 10, hex = base 16, etc. 13:24:11 Just do the mod operation? 13:24:20 Yes, mod by the base, divide by the base, repeat. 13:24:29 --- join: Pepe_Le_Pew (n=User@201008242045.user.veloxzone.com.br) joined #forth 13:24:31 Use the appropriate alphabet of symbols. 13:25:03 In Forth, this is what the <# # #S #> words to do produce numeric output for any value of BASE {2..36}. 13:25:22 eeew!!! 13:25:33 Is that such a good idea? 13:25:37 huh? 13:25:55 Are stacks able to do that kind of operation? 13:25:59 Of course they are. 13:26:03 I thought they can only add numbers? 13:26:07