00:00:00 --- log: started forth/05.09.22 00:01:37 --- join: amca (n=plump@as-bri-1-223.ozonline.com.au) joined #forth 01:56:58 --- join: amca_ (n=plump@as-bri-1-197.ozonline.com.au) joined #forth 01:58:32 --- quit: amca (Read error: 110 (Connection timed out)) 02:25:06 --- join: tattrdkat (n=virsys@or-65-40-177-69.dyn.sprint-hsd.net) joined #forth 02:25:09 --- quit: virsys (Read error: 110 (Connection timed out)) 03:28:29 --- part: amca_ left #forth 03:41:45 --- quit: Quartus (Remote closed the connection) 05:43:24 --- nick: Raystm2 -> nanstm 05:51:10 --- join: PoppaVic (n=pete@0-1pool73-101.nas24.chicago4.il.us.da.qwest.net) joined #forth 05:51:32 Mornin' 05:52:54 --- join: aardvarx (n=folajimi@shell2.sea5.speakeasy.net) joined #forth 05:56:39 Morning, all. 05:56:46 howdy 05:57:03 Not bad, not bad. 05:57:08 Thank you for asking. 05:57:11 How about you? 05:57:27 Fine. Wakin' up. 06:06:39 --- quit: crc (Read error: 110 (Connection timed out)) 06:11:01 --- join: madwork (n=foo@derby.metrics.com) joined #forth 06:19:34 good morn 06:19:49 hiya 06:22:05 good conversations, the last few day. 06:22:32 Glad to help ;-) It seems to aggravate some folks. 06:23:16 Live development, new entertainment. 06:24:03 I've further simplified the std "opcodes", but I'm also thinkin' about the std branch/flow words... I think they are higher than this level. 06:28:44 Not sure I can really do what I need... Trying to think up & down code and nesting and shit can be a real bother. 06:29:42 The excersize is not lost on me. 06:29:48 heh. 06:30:04 ballance away. 06:31:35 OK, gimme' some insights here... I've added opcodes for assorted jmps (aka "branching" - such as jmp, zjmp, nzjmp)... These are the lowest WE suffer - at the vm. These are NOT complete "begin...repeat" or "if...then". 06:32:49 I'm just about convinced that we need to look at the vm "registers" or RS value, and tweak from there to really get the familiar flow-controls. 06:34:04 OR.. Maybe I am being silly, and I should just write those familiar flow-controls as opcodes, and dump the more universal 'jmp' stuff? 06:34:16 * Ray_work novice and not much help, but please continue to think aloud so I can get an appreciation for the process. 06:35:01 http://rafb.net/paste/results/4C3Lof85.html 06:35:18 Above is the streamlined "opcodes". See the stuff around C(jmp) 06:35:40 line 32+ 06:35:43 clicking... 06:36:09 Down to 93 opcodes in this table, anyway 06:39:36 As you can see, this table/opcode-set is about the bare-bones minimum for just about anything (excepting strings and headers/vocs). 06:40:18 or, strings+io, headers+vocs. 06:40:50 _mechanically_, these opcodes should let us DO just about anything, at the bytecode level. 06:42:20 But, I find myself glaring at the 'jmp' stuff and pondering. Readability would be OK for asm-familiar folks, I suspect. But, decompiling would suffer nastily. 06:44:56 --- join: sproingie (i=foobar@64-121-2-59.c3-0.sfrn-ubr8.sfrn.ca.cable.rcn.com) joined #forth 06:46:20 --- join: JasonWoof (n=jason@pdpc/supporter/student/Herkamire) joined #forth 06:46:20 --- mode: ChanServ set +o JasonWoof 06:46:26 lo, jas 06:46:56 yo 06:47:31 --- join: Quartus (n=trailer@ansuz.pair.com) joined #forth 06:47:34 Hey all. 06:48:47 hi Q 06:49:26 hey, jas? YOu've a running herkforth there, right? Using the opcodes? Do you have a decompiler ala' "see" running, too? 06:53:52 hmm 06:54:04 I think JasonWoof fell asleep 06:54:36 Decompilation is highly over-rated. 06:54:53 Not in an interactive environment it isn't. 06:55:01 Decompilation in any enviroment is highly over-rated. 06:55:22 I will definitely agree that I remember vfa's and fpc/f83 fondly, however. 06:57:13 Why decompile? 1) Out of curiousity 2) ...? To see that the compiler is working properly? 06:57:38 Both. Examining what already works, and checking you laid down the proper code. 06:58:16 If you need to check the decompilation to see if you're coding right... do I need to finish that sentence? 06:58:52 no, and you'd be wrong 06:59:35 Then you're incapable of coding in a language that doesn't provide decompilation facilities? 06:59:36 I've had to do this several times just today - because there are times when words COMPILE, but they are just not right... using DOES> can do this. 07:00:00 Quartus: don't be silly, I've coded far longer in C than I ever did in forth. 07:00:08 Which can also be decompiled. 07:00:38 Ok, so you have a sketchy understanding of when to use what compilation word, so decompilation helps you figure out what you're doing? 07:00:44 no, it may be unassembled, but the chances of "decompiling C" are seriously limited. 07:00:51 Right 07:01:08 if the code laid down looks "wrong", it means you missed some tidbit 07:01:14 Then I clearly do need to finish that sentence -- If you need to check the decompilation to see if you're coding right, you don't understand the programming language well enough yet. 07:01:51 Quartus: forth has changed greatly in 20 years, and there are so many variants that I do indeed like to have "see" around. 07:02:27 I could certainly live w/o it. But, then - I can live w/o a lot of things. 07:02:31 That doesn't alter the truth of what I've just said -- it means you don't understand the particular variant that you're using well enough. 07:03:12 Quartus: and you'd be correct, since it has appeared for years that every variant strives to out-variant the prior variant. 07:03:25 Though decompilation will usually represent the underlying code in a format similar to the source, so if you don't understand the semantics of the language (or variant) properly, you'd be no further ahead looking at the decompilation. 07:03:37 It may also have a lot to do with docs that often say something like "to be written" 07:04:24 Anyway, this takes us quite far afield. Frankly, reintroducing VFA's is prolly more than sufficient. 07:04:31 What is a VFA? 07:04:37 view-field-address 07:04:44 What is a view-field-address? 07:05:00 oop 07:05:01 back 07:05:10 I don't decompile 07:05:12 don't see the point in it 07:05:16 a header-component that was used by F83 and FPC to get "help" by just listing the block or source/lines of the "definition" 07:05:16 I'd rather look at the source 07:05:20 hi JasonWoof hi Quartus. A good day to you all. 07:05:26 Hi Ray. 07:05:29 hi ray 07:05:38 pray we don't wash away this weekend... 07:05:39 JasonWoof: cool. That settled that, then ;-) 07:05:51 herkforth shows you the source all the time 07:06:08 Keeping fingers crossed, Ray. 07:06:12 I can't think of any reason it'd be better to look at dissassembly than the source 07:06:18 unless you don't trust the compiler 07:06:19 Ray_work: well, Rita is likely to start a lot of folks thinking about hand-holding-government, anyway ;-) 07:06:42 my herkforth environment is not yet rewritten for fovium 07:06:45 JasonWoof: I agree all the way, and yeah: I think the VFA solutions are the way I'll go 07:06:46 I'm working on it though 07:06:58 SEE is great stuff on some systems 07:07:02 a self-documenting forth system would be awesome 07:07:11 JasonWoof: you now need to argue "see" with Q ;-) 07:07:13 but mostly because it's the only convenient way to find out how words work (or even what they do) 07:07:16 and i don't mean blocks 07:07:18 right 07:07:33 if there's an easier/better way to do it like VIEW, then see is useless 07:07:35 sproingie: I agree, I'm thinking - I'm thinking! 07:07:40 tho i think it would annoy most people to have to write documentation longer than the word 07:07:46 hi sproingie :) 07:07:50 hey all 07:08:02 sproingie: that's true in every language, and it's still necessary 07:08:03 I'm in favour of literate programming for large projects. 07:08:10 can't force people to write legible code :) 07:08:16 right 07:08:25 more true in forth -- i suppose top-down documentation would work out 07:08:40 Quartus Forth itself is written that way; it generates as a document or as an executable. 07:08:43 document the high level words, then as time and inclination permit, go to the low level words 07:09:08 Quartus: sounds like literate programming. i'm thinking more like the way emacs does it 07:09:16 sproingie, what are you talking about? 07:09:24 or the way python could have done it if it weren't so bad at itt 07:09:36 My only bitch about documenting funcs, structs and headers is: 1) Either you must do almost the whole header again PLUS or; 2) you use a freaky tool to generate some silly thing that might almost help. 07:09:42 Quartus, why do I still need to put comments in code if the code can "document itself?" 07:09:59 aardvarx, code can only document how it does what it does -- not why. 07:10:00 aardvarx: self-documenting systems. where everything can have a docstring -- functions, variables, whatever. combined with a really good hypertextified search interface 07:10:08 aardvarx: better code and languages simply reduce the AMOUNT of comments needed. 07:10:32 sproingie: wtf is a "docstring"? 07:10:49 Quartus: i disagree somewhat -- it should also mention just what it does, in some detail, if it's a high level word 07:10:58 right 07:11:01 Quartus: rather than force you to drill down to each low level piece 07:11:15 sproingie, you disagree with what? That code cannot document its own 'why'? 07:11:15 otherwise i agree, docs should be about why, not how 07:11:41 For example, say I want to use a 'for loop' in C as opposed to a while loop; I then say "/* for loop used here to avoid infinite looping that is possible with a while loop */ 07:11:42 Quartus: that it's sometimes not obvious when it's really abstract 07:11:51 in forth, for example: 1) typify input and output; 2) definitely mention any vars/stacks that get altered such that some other word can blow up 07:11:54 Are you saying that the code can do that by itself? 07:11:55 sproingie, that sounds like agreement, not disagreement. 07:12:13 Quartus: qualified agreement anyway 07:12:18 aardvarx: the for/while comment sounds bad 07:12:21 mostly "why" and a bit of "what" 07:12:27 right 07:12:28 PoppaVic: in langs like python, a docstring is a string in an agreed apon position in the def, where an application can scour your work and just pick up all the names and docstrings and auto document your work. 07:12:37 PoppaVic, could you please explain? 07:12:37 ahhhhh 07:12:57 PoppaVic: just verbose. i might put "the obvious 'while' loop is infinite -- need the pre-test of 'for'" or whatnot. or just fix the code instead :) 07:12:57 Ray_work: a verbous vfa! 07:13:09 sproingie: right 07:13:12 A stack diagram and a comment are the 'what' for a Forth word. The 'why' is very rarely going to be self-evident. The 'big-picture' needs docs, too. 07:13:18 PoppaVic, was my line of reasoning wrong? 07:13:50 aardvarx, that would be over-commenting in my book. Comments shouldn't explain the programming language. 07:13:52 a lot of emacs is representative of very excellent documentation 07:13:52 aardvarx: there are times for for() and while() and even do{}while() - the code should strike you as the best solution to 2+ issues, or it just doesn't matter. 07:14:23 swap \ swaps the top two stack items ( a b -- b a ) is not good commenting. 07:14:32 * aardvarx shocked by Quartus analysis 07:14:43 Quartus: actually it makes a great docstring 07:14:44 Quartus, how can I learn to comment properly? 07:15:25 aardvarx: read "Code Complete". it's by microsoft press, but don't let that turn you off 07:15:34 PoppaVic, I do understand what you are saying -- "Use the right tool for each job", right? 07:15:48 * aardvarx googles for sproingie's suggestion... 07:16:18 Straightforward rules-of-thumb; assume the reader is conversant with the programming language. Explain why. Explain how only when it's necessary. Write to a larger context -- "a++; /* increment a by one */' is not good commenting. Don't write arcane or opaque code in the first place, and if you have to, explain it thoroughly. 07:16:25 goddamnit's getting dark 07:16:32 Code Complete is actually a very good book, yes. 07:16:32 microsoft press is just a publisher. their books tend to be top notch 07:17:09 i liked Don Box's Inside COM too ... tho i dont think that was MS press 07:17:15 not relevant to style anyway :) 07:17:18 Quartus: decent thumb-rules 07:17:23 aardvarx, also -- use descriptive variable and function names. 07:17:29 gods yes 07:17:41 Quartus, I did learn that in school. 07:17:53 Don't go overboard; Hungarian notation is ridiculous, for instance. Use i for a loop index, unless there's some really compelling reason not to. 07:17:54 What I did not learn is everything else you mentioned :( 07:17:54 there's definitely a difference between good comments and good docstrings 07:18:01 and, (and this from my early forth years): try to keep functions as FUNCTIONS and always within a screen 07:18:16 docstrings should usually assume less familiarity with the language, they're aimed at people learning the API, not necessarily the ins and outs of the language 07:18:21 Yes, factor well. 07:18:22 sproingie, comments != docstrings ? 07:18:43 aardvarx: the latter means, if a func starts hitting 2_ loops or 40+ lines, you need to break it down more. 07:19:00 I see. 07:19:08 aardvarx: docstrings are a synopsis sort of comment at the very top, written as a string. lisp and python treat them specially 07:19:14 Assume familiarity with the idioms of the language, too. 07:19:26 aardvarx: from what sproingie and ray have said, no.. comments are immediate notes. docstrings are like precursors to manpage-entries 07:19:33 emacs has oodles of good docstrings. python's are quite a mixed bag 07:19:48 Quartus, do you have any examples to distinguish between good and bad commenting? 07:20:03 PoppaVic, sorry to put you on the spot, but... 07:20:16 I'd rather hear of good/bad "docstrings" and how they'd "compile" 07:20:22 aardvarx: np 07:20:27 PoppaVic: heh if your forth word is "within a screen", that doesn't say much 07:20:29 do you have code samples which exemplify proper commenting? 07:20:54 sproingie: forth, Cm whatever - I try to keep shit short & sweet. 07:20:58 Easier to find the ones that aren't good, aardvarx :) 07:21:06 Hmm 07:21:16 Bummer. 07:21:22 most BAD comments are those kids throw off just "because the teacher said so" 07:21:25 i was reading somewhere where one person was asked on an interview question, "how long should a method be, at most?" 07:21:35 sproingie: right 07:21:42 he replied "no bigger than my head". the interview asked "so, no bigger than you can keep in your head at one time?" 07:21:51 Code Complete has some good examples. So does "The Practice of Programming", and "Writing Solid Code". 07:21:58 he responded "no, i put my head up to the monitor. no bigger than that" 07:22:24 I'm a bit leery of Code Complete; it looks like it only uses MS tools. :( 07:22:28 short & sweet is best, and yeah the screen-size. For most folks, we think 80x24. 07:22:35 aardvarx: if only MS followed the guidelines in CC 07:22:45 aardvarx, it doesn't use any tools -- it's programming philosophy. 07:22:58 "Thinking Forth" has discussion of comments too, I believe; freely available on the weg. 07:23:01 web, even. 07:23:04 hell if only MS followed their own UI guidelines 07:23:09 Brodies "Thinking Forth" was pretty handy as well 07:23:27 I have to dig out my copy, soon 07:23:49 one of those guidelines is, in a nutshell, "avoid cascading menus. if you must use them, make them no more than one level deep off the main menu" 07:23:54 then they give us the start button 07:23:54 sproingie: no doubt - and we have M$ versus Apple versus GNU/gnome 07:23:57 "The elements of programming style" is another oldie but goodie. 07:24:15 yes, I recall it vaguely,Q 07:24:27 thinking forth is really quite nice 07:24:34 Quartus, the other two references are very helpful, though :) 07:24:42 sproingie: I need to check his DOER code again, too 07:24:49 i wasn't a big fan of starting forth, it runs out of steam and goes straight for the guts of a forth interpreter 07:24:58 Code Complete is, again, quite good, and does not focus on MS tools. 07:25:06 sproingie: worse, it was FORTH, INC targeted 07:25:11 and the analogy it uses for CREATE..DOES> is stretched and hard to follow 07:25:35 I believe I may return to the paradigm 07:25:36 thinking forth is pretty good 07:25:38 sproingie -- I agree with you. So many books start to teach Forth, and then launch into the structure of the dictionary header, and threaded internals, etc. 07:25:58 I was really unhappy to see many tests COMPILE that lacked create, but had does> 07:26:58 there should be a starting forth type doc for gforth and retroforth 07:27:22 Anyway, that's every book I own that I'd recommend for good coding practices. 07:27:22 Unless, of course, I've missed something about DOES> that can skip a generation. 07:27:33 sproingie: I agree 07:27:33 rf's perhaps too much in a bit of flux to finalize a starting forth for it, but some stuff is solid, like its unusual way of handling conditionals 07:27:37 'skip a generation'? 07:28:11 one thing i'd leave out is the nasty fixed-point stuff. this is the 21st century, we have insane fast dedicated hardware for floating point 07:28:15 Quartus: oh. Thinking maybe you'd define a does> word early, use it later on with a word using create 07:28:26 PoppaVic, you can do that. 07:28:32 sproingie: I only partially agree 07:28:52 Quartus: ahh, then this is why they dropped for create does> ? 07:28:54 --- join: thinfu (i=thin@bespin.org) joined #forth 07:28:54 --- mode: ChanServ set +o thinfu 07:29:17 Yes, floating-point is sometimes very useful, but can cause disasterous problems if you are unaware of its limitations. 07:29:23 sproingie: there are a ton of situations where a single works great. 07:29:27 PoppaVic, yes. 07:29:32 rf only runs on i386, so people learning to use forth on an ARM or whatnot are going to have to learn fixed point elsewhere 07:29:40 Yes to the 'why they dropped . 07:30:06 this is why I merged the single-precision with "core" and made the double (long) and reals (floats) optional. 07:30:31 Quartus: that explains a lot. Too bad it's not clear from docs I have here. bastards! ;-) 07:30:41 single may work for maths, but i'm thinking of a tutorial on writing pixel shaders in forth 07:30:47 once i learn shaders myself 07:31:14 basically, something relevant. no one cares to learn how to print lines of stars to their terminal 07:31:16 sproingie: you could explain 3D wireframes and viewpoint as well 07:31:47 But I agree, treating the screen as more simple-minded than a teletye was depressing. 07:31:53 teletype 07:32:48 http://rafb.net/paste/results/k7C0ZC21.html 07:33:16 Q, sp: above is the latest opcore.h - I think that covers the eseentials, sans io/strings 07:33:35 Going to update the wiki? 07:34:07 Quartus: I'm cogitating that - would it be better to update the LINKS to other pages, or just add new linkages? 07:34:52 Up to you, I guess. 07:35:13 I may also use an "attach" - assuming a tarball would work. (ie: I can upload that tarball) 07:35:19 Sure. 07:35:27 Plenty of room on the server. 07:35:42 Cool, I did not want to be a pig or sloppy 07:35:51 Go nuts. 07:35:55 --- part: thinfu left #forth 07:36:01 you-da-man! ;-) 07:36:06 * JasonWoof wanders back in 07:36:35 I think the ease in factoring forth can help a great deal in making your code easy to understand 07:37:01 yep 07:37:06 I've sometimes seen over-factoring, but usually the opposite. 07:37:14 This is also true of using structs, locals and funcs in C 07:37:15 but I think it is very very important to have a browser/editor/interpreter/whatever environment where you can easily see the definitions of things 07:37:49 if you have to grep around for definitions, then it becomes frustrating when things are highly factored 07:37:53 Quartus: well, my own "rule-of-thumb" is: will/can I reuse this? 07:38:05 Well, sure, PoppaVic. But I've seen : a! a ! ; 07:38:23 Quartus: yeah, me too - damned silly 07:38:55 the only excuse can be that it is used soooooo often, you save a shitload of space 07:39:09 Idioms do not need to be factored, unless so doing hides an implementation detail. 07:39:22 And if it does that, it needs to be well-named. 07:39:25 Quartus: yeah, I've done stuff like that probably. now I realize that taking a space out is not useful 07:39:31 This is why I think we could benefit from a "true macro" 07:39:48 I had a Quartus Forth user ask me to day why 1+! wasn't in the kernel. 07:39:54 having something named only makes sense if it's something nameably (in that case you're just typing the contents) 07:40:01 how would a "true macro" function in forth? 07:40:20 : a! a ! ; could be useful if it was for something you might change how it is set later 07:40:27 forth is driven by the lexer ... something like a lisp macro would be ... difficult 07:40:31 JasonWoof, hopefully it's not called a!, in that case! :) 07:40:46 why not? 07:40:54 Because that says nothing about what it does. 07:41:02 to me a! means "set a" 07:41:21 this being an example, presumably a is metasyntactic .. 07:41:23 eg yesterday I implemented here! 07:41:33 If it sets a limit, call it 'set-limit'... the fact that it stores a value in a variable isn't useful information at the next level of abstraction. And a name should advance the level of abstraction. 07:41:51 to me, foo! means do foo with some side effect 07:41:57 Quartus, speaking of kernels... 07:41:59 but that's just the schemer in me 07:42:03 are you familiar with L4 ? 07:42:08 aardvarx, no. 07:42:10 sproingie: a "true macro" would just be a stored string, and the func to shovel it into the input-stream in PLACE of some other text. 07:42:17 sproingie: yeah, ! means "store" in forth 07:42:21 PoppaVic, Wil Baden has written such a facility. 07:42:32 PoppaVic: not very interesting unless it takes parameters 07:42:40 Wil's does. 07:42:41 Quartus: yep, my ancient Forth Dimensions mentioned them 07:42:50 Quartus: yeah, we have different naming conventions. I consider "!" a good shorthand for "set-" 07:42:57 not an exact match, but close enough 07:42:58 sproingie: perhaps, THAT I'd like to see 07:43:03 how does it take parameters? forth style? 07:43:07 hm... actually 07:43:18 that's not my convention at all... usually I'd name it >here 07:43:20 i mean you could run forth code through a macro preprocessor, but ick 07:43:28 yeah 07:43:29 * JasonWoof renames a word now 07:43:44 sproingie, it replaces markers in the macro string with parsed text, as I recall. 07:43:47 to me, >foo means push onto the foo stack 07:43:52 where !foo means set foo 07:43:56 and @foo means get foo 07:44:03 well, perhaps we could use a "pre-interpreter" - I dunno 07:44:14 JasonWoof, >foo means to either move, or translate. For translation words I prefer before>after. 07:44:19 foo@ does a foo-fetch 07:44:22 To me, anyway :) 07:44:23 yeah 07:44:28 I do > for putting something somewhere 07:44:28 Quartus: agreed 07:44:31 -> for convert 07:44:39 dict->cfa 07:44:40 or at least a Cish foo2bar 07:44:48 dict->cfa ( dict -- cfa ) 07:44:51 foo@ to me means fetch the foo at that address 07:45:00 ah pity da foo 07:45:07 : >a ( x -- ) \ put's TOS in the A register 07:45:10 Quartus Forth, being 16-bit, has >rel and >abs, which do translation; they're not ideally named, but it's a bit late to rename them. 07:45:36 sproingie: if foo! puts to foo; foo@ fetches it. It's not that difficult. 07:45:41 I like ">here" much better than "here!". because it does something very different from "here !" 07:45:50 Quartus: not a lot of options there 07:45:59 PoppaVic: fits your convention .. i just move the ! and @ to the left for mine 07:46:01 PoppaVic, abs>rel and rel>abs, perhaps. 07:46:12 hmm 07:46:19 yeah, Quartus that's better 07:46:21 yeah, I have the rather nice freedom to rename anything easily and instantly and without bothering anyone 07:46:22 I went with the shorter version, because it's on a Palm. 07:46:25 tho the only real code i have recently has even odder naming 07:46:31 I still think 'typing' the stack would benefit us. 07:46:57 Quartus: when space is limited, cheat however you can 07:47:02 char+@ and char+! ... but those are setting at absolute addresses, not getting variables 07:47:06 PoppaVic: typing the stack? what do you mean by that? 07:47:15 the + means it also advances by one (variable sized) char 07:47:30 sproingie -- char+@ is COUNT, right? 07:47:50 JasonWoof: an idea we chucked around a few days back.. I think it was Quartus that referenced strongforth, but something similar - only for COMPILING 07:48:09 Not me. I've heard about strongforth, but the idea of a typed Forth makes me itch all over. 07:48:18 hmm 07:48:24 maybe sproingie then 07:48:45 Quartus: char+@ ( a -- a' u ) | returns codepoint at a, returns address of the next char and codepoint 07:48:52 Quartus: not the language - just while interpretive-compiling. 07:48:58 sproingie -- codepoint? 07:49:00 whups, bad comment. wrote that on the fly 07:49:05 oh data types 07:49:05 ok 07:49:08 nm 07:49:09 Quartus: integer value of a unicode char 07:49:13 * JasonWoof wanders of to find food 07:49:25 it's half of the utf-8 codec 07:49:25 sproingie, then yes, traditionally called COUNT. 07:49:30 JasonWoof: yeah, typing stack-entries broadly. 07:49:41 Quartus: what's this about COUNT? 07:49:54 did i reinvent the wheel here? 07:49:59 yep 07:50:02 Standard word. COUNT ( c-addr1 -- c-addr2 c ) 07:50:08 I'd use an ALIAS 07:50:09 c-addr2 is one char further along than c-addr1. 07:50:26 i don't recall seeing COUNT in DPANS 07:50:32 It's there. In CORE. 07:50:46 oy. i'll check 07:50:52 http://quartus.net/dpans/dpans6.htm#6.1.0980 07:50:56 I still need to create an opset for strings. 07:51:49 PoppaVic: ah, that returns an entire string. this returns the char itself 07:52:01 hmm???? 07:52:02 maybe i could work COUNT into the naming scheme though 07:52:04 no 07:52:17 count can be looped to index the string and get each char 07:52:18 sproingie - COUNT? No, COUNT does exactly what your char+@ does. 07:52:46 The ( c-addr u ) isn't a string, as such. It's an incremented address, and a character. 07:52:50 not at all. it returns the next address all right, but the u is the "size in characters 07:52:56 Nope. 07:52:58 i'm not interested in the size, i'm interested in the char itself 07:53:02 UNLESS you are redefining the count-element as 16 or 32 bit 07:53:15 sproingie, it's only the count if it's a counted string. 07:53:51 hm. i should rename it utf8-count then 07:54:08 is there an analogous word for setting the char at the address? 07:54:12 If your char size is set appropriately, COUNT will work. 07:54:46 tho setting creates all sorts of other happy fun insanity, like replacing a 3-byte char with a single byte 07:55:04 even utf-16 doesn't help, thanks to surrogates 07:55:20 sproingie, no counterpart for saving exists. What is your stack diagram for the counterpart? 07:55:28 Quartus: the whole point of utf8 is that there is no single character size 07:55:35 ummm 07:55:37 no 07:55:56 utf-8 has characters that are from 1 to 4 bytes long 07:56:02 the point to utf8 is to merge unishit and ascii 07:56:23 the point of it is to be backward compatible with low ascii, yes 07:56:46 it's a lousy encoding for chinese, but it seems to have taken over nonetheless 07:56:50 I certainly will need a converter, but I'm already planning for utf8 in files, and wchar_t in memory. 07:57:40 everyone bitches about their chinese or korean - it's old and tedious. 07:57:42 i also plan on a utf-16 and a ucs2 codec. ucs2 is easy, no surrogates, but it's BMP only 07:57:57 in korea only old people speak korean :p 07:58:36 I figure ascii rules, and forth is the "one ring" - so adding wchar_t in memory is about as polite as you can get. 07:58:39 I would think for sanity you'd want to pad out a utf8 character to four bytes regardless. 07:58:40 anyway, i gotta figure out some way to handle botched decoding before proceeding any further 07:58:58 Quartus: i can't very well require everyone who passes a ut8 string to "pad it out" 07:59:03 I mean internally. 07:59:05 the point of a codec is to read what you're given 07:59:05 On the stack. 07:59:16 Quartus: I just KNOW you don't mean in file - oh,nm 07:59:25 sproingie, yes, I'm faintly familiar with the point of encoding schemes. 07:59:25 Quartus: that's exactly what the codec does, it converts it to 32 bit ints one char at a time 08:00:07 Sadly, wchar_t and unishit sorta' ruin the idea of indexing forward and back nicely & neatly. 08:00:10 i'll probably also write a utf-8<->utf-16 codec as well 08:00:38 sproingie: luckily, C has wchar_t alreay ;-> 08:00:42 already, two 08:00:44 PoppaVic: blame the fact that bytes are only 8 bits. roman was never the only character set in the world 08:00:44 PoppaVic, you have to stop mangling names of things you don't like by making the last syllable 'shit'. It makes it very hard to take you seriously. 08:01:20 PoppaVic: string operations were *never* an easy problem in non-english locales. unicode ain't easy, but it's *vastly* easier than the mess that came before 08:01:42 strings are hard. ascii just happens to be a really simple degenerate subcase of string handling 08:01:43 Quartus: unifoo, then - whatever floats your boat. When you've got more restrictions and rules than working libs that are usuable, it's hard for me to worry much. 08:01:58 sproingie: exactly 08:02:23 There's a character on a show up here-- "Trailer Park Boys" -- the park supervisor, Lahey, does that, and it's hard not to think I'm hearing Lahey speak when you write like that. It's amusing, but it doesn't lend any credibility. 08:02:24 sproingie: the only "plus" I can see for wchar_t is that they support the whole schlock. 08:03:16