00:00:00 --- log: started forth/02.08.21 00:07:50 --- quit: yeahright (Read error: 110 (Connection timed out)) 00:08:12 --- join: yeahright (~proteusgu@24-197-147-197.charterga.net) joined #forth 00:42:52 --- quit: Speuler (Read error: 110 (Connection timed out)) 01:07:59 --- join: proteus (~proteusgu@24-197-147-197.charterga.net) joined #forth 01:08:18 --- quit: futhin (Read error: 113 (No route to host)) 01:09:17 --- quit: yeahright (Read error: 110 (Connection timed out)) 02:07:34 --- quit: proteus (Read error: 110 (Connection timed out)) 02:07:50 --- join: proteus (~proteusgu@24-197-147-197.charterga.net) joined #forth 02:41:04 --- quit: Soap` (Read error: 104 (Connection reset by peer)) 03:07:43 --- join: yeahright (~proteusgu@24-197-147-197.charterga.net) joined #forth 03:07:44 --- quit: proteus (Read error: 110 (Connection timed out)) 03:51:52 --- quit: pyromaniac (Connection timed out) 03:55:54 --- join: pyromaniac (~pyromania@dialup-217.159.220.203.acc01-high-pen.comindico.com.au) joined #forth 04:07:45 --- quit: yeahright (Read error: 110 (Connection timed out)) 04:07:54 --- join: yeahright (~proteusgu@24-197-147-197.charterga.net) joined #forth 04:48:45 --- join: Serg_penguin (~Z@nat-ch1.nat.comex.ru) joined #forth 04:49:27 --- quit: Serg_penguin (Client Quit) 04:58:18 --- join: sif (~sifforth@ip68-9-70-120.ri.ri.cox.net) joined #forth 04:58:18 Type sif: (or /msg sif to play in private) 05:07:26 --- quit: yeahright (Read error: 110 (Connection timed out)) 05:07:54 --- join: yeahright (~proteusgu@24-197-147-197.charterga.net) joined #forth 05:26:01 --- join: Fare (~fare@grenoble-1-a7-62-147-72-91.dial.proxad.net) joined #forth 05:45:28 --- join: cleverdra (julianf@0-2pool111-57.nas2.florence1.sc.us.da.qwest.net) joined #forth 06:06:51 --- quit: yeahright (Read error: 110 (Connection timed out)) 06:07:04 --- join: yeahright (~proteusgu@24-197-147-197.charterga.net) joined #forth 06:08:42 --- join: tathi (~josh@wsip68-15-54-54.ri.ri.cox.net) joined #forth 06:09:55 tathi :) 06:11:25 hey 06:17:27 --- nick: yeahright -> ProteusGuy 06:25:23 --- join: Serg_penguin (~Z@nat-ch1.nat.comex.ru) joined #forth 06:25:41 hi 06:27:55 --- quit: Serg_penguin (Client Quit) 06:39:36 --- quit: Fare (Read error: 104 (Connection reset by peer)) 06:42:43 hello tathi 06:42:56 i dont c u 2 much nowdays 06:43:50 the channel is very active these days i think 06:44:09 so its worth 2 b here 06:44:41 how is ur 68k4th? 06:45:07 managed 2 write drivers? 06:46:02 onetom: I've been on vacation for two weeks 06:47:02 and PPC 4th, not 68K. It's coming along, I keep rewriting it to make it simpler, so I haven't gotten all that far yet :) 06:48:48 think I've gotten the basic engine simplified as far as it's going to go 06:49:13 now I'm fixing all my low-level forth code to run with the engine changes that I've made 06:49:24 --- join: flyfly (~marekb@217.66.164.22) joined #forth 07:07:25 --- quit: ProteusGuy (Read error: 110 (Connection timed out)) 07:08:02 --- join: ProteusGuy (~proteusgu@24-197-147-197.charterga.net) joined #forth 07:59:50 aint the ppc is a 68k derivative? 08:00:37 * onetom damn sleepy coz just woke up (the phone rung) 08:02:07 not really 08:02:20 closer to MIPS, that kind of thing, I think 08:03:37 ah. good 2 know. they always confuse me 08:03:57 and can u say about its performance? 08:04:16 last time u were stucked w the various caches 08:04:28 --- quit: cleverdra (Read error: 104 (Connection reset by peer)) 08:04:40 yeah, not great for self-modifying code 08:04:53 not really sure about performance 08:05:20 way shorter pipeline than Pentiums, which helps 08:05:54 think the floating point & vector stuff is pretty fast, but integer performance isn't terrific 08:06:25 but I haven't actually done any tests, that's just impressions I've gathered 08:06:57 --- join: cleverdra (julianf@0-2pool111-57.nas2.florence1.sc.us.da.qwest.net) joined #forth 08:07:13 --- quit: ProteusGuy (Read error: 110 (Connection timed out)) 08:07:14 Dan Bernstein (http://cr.yp.to/) makes it sound like it's kind of slow, but think he's doing cryptography stuff so mostly using the integer units 08:07:19 --- join: ProteusGuy (~proteusgu@24-197-147-197.charterga.net) joined #forth 08:07:55 (eww, slow integers?) 08:08:49 tathi: ? 08:08:55 seems that way, think it was designed partly for scientific applications or something 08:09:56 hmm sounds really odd... 08:10:33 cleverdra: PPC isn't exactly suited to forth; much better for C with an optimizing compiler :) :( 08:11:37 well, I'll take it over x86 any day 08:13:00 fixed length instructions makes writing assemblers and disassemblers almost trivial, for one thing 08:14:25 thing I really like is the way apple makes dual-processor machines 08:15:20 I can compile something huge on one processor and play games or whatever on the other one while I'm waiting :) 08:22:01 hmmmm, yummm :) 08:22:21 how fast is it compared 2 the x86 product line? 08:22:48 that's what I don't know 08:24:02 I _think_ it's more instructions per clock speed, and the pipeline is shorter, so bad instruction scheduling isn't as much of a speed hit as x86 08:24:03 :)) 08:24:44 cant we choose an app existing for both machine? 08:25:01 do u use its "native" OS or linux? 08:25:06 ah, not linux 08:25:18 uve told it already afair 08:25:27 ? 08:25:33 I'm running linux 08:25:41 wow. debian? 08:25:47 I have MacOS 9 installed, but I almost never run it 08:25:56 No, built it myself 08:26:03 c 08:26:28 Herk is using Gentoo, I might switch to that at some point 08:26:30 then, we can easily make some test. eg, some image manipulation w imagemagick 08:26:57 yeah, probably 08:27:00 or how much does it take 2 compile a linux kernel? 08:27:34 hmm, didn't time that 08:27:46 probably about time to build a new one though... :) 08:28:12 ? that is? 08:29:20 you know how long it takes to compile a linux kernel on your machine? 08:30:05 ~20mins a full general kernel w all the modules on a PII-600 1CPU 08:30:40 i do divx film recompressions w MEncoder 08:31:06 you do what? 08:31:12 that works 12fps @ ~1000kbit/s 08:31:34 that is half-realtime 08:32:11 i often got films longer than b able 2 fit on a 80mins cd 08:32:33 so i have 2 recompress them w a slightly lower bitrate 08:32:37 c now? 08:32:40 yeah 08:33:22 I just really don't care that much about speed, since I don't do anything processor intensive 08:34:00 herk might have a better idea of how fast the thing is, as he's into fractals 08:35:08 so u r absolutely satisfied w its performance, right? 08:35:26 the wordprocessor, the browser isnt slow @ all? 08:35:32 do u use mozilla, eg? 08:36:07 mozilla is slow on anything :) 08:36:27 I'm not running that at home 08:36:30 lollllllll 08:36:34 okay 08:36:50 btw, have tried dillo? 08:36:52 takes about 5 seconds to start up on my laptop at work, which is a PowerBook running a 500 MHz G3 08:36:54 btw, have u tried dillo? 08:36:57 after it starts up it's OK 08:37:12 yeah, I use lynx and dillo for practically all my web surfing 08:37:35 hacked it a bit so I can go forward, back, and reload from the keyboard though 08:37:48 (one of those didn't have a key bound to it, can't remember which) 08:38:25 umean in dillo? 08:38:35 yes 08:38:41 yeah, thats 2 young 08:39:47 seemed kind of weird that they were all into getting cookie support and stuff, but hadn't added key bindings for everything yet 08:40:43 OK, that's right around 10 minutes for a kernel build 08:40:56 that's my standard config though, so it doesn't include a bunch of the modules that I have no use for 08:41:16 on my 533 MHz G4 at home 08:41:58 without doing make -j 2 or anything, so using just one processor 08:47:01 aha, so the performance/cpu-freq is similar on both machines... 08:47:26 imean there is no >2x factor between them 08:47:34 yeah 08:47:58 gcc for PPC produces suboptimal code. I had that problem when working on 850 and 860 powerQUIIC at my last job. 08:48:25 i would b very interested in the performance of a 500MHz 4th processor.... 08:48:36 I know, I've looked at some of the output 08:48:40 the biggest problem was gcc was wrong on a lot of it's compile-time branch prediction. x86 uses runtime branch prediction by having like twice as many pipelines. 08:48:48 hiii! 08:48:56 hi:) 08:49:54 it sucks on 08:50:09 MIPS too. cuz the compiler has to manage the cache itself 08:51:10 yeah 08:51:35 the biggest problem with a stack-based processor is trying to keep stack accesses fast. it's easy to keep register access fast on a traditional processor and just have memory access be slow. 08:52:05 you'd definently want to make a stack processor a harvard architecture. so you can do stack access and load instructions at the same time. 08:52:37 --- quit: Stepan ("Do you think it is air you are breathing? Hmm?") 08:53:03 OrngeTide: why do u state so solid that keeping stack access fast is so problematic? 08:53:07 or like mup21 where you can load like 5 instructions at a time. :) 08:53:57 onetom, because i'm assuming you want to support an arbitrarily large stack. which couldn't exist in fast flip-flop registers you'd have to put it into addressable memory (dram or sram) 08:54:11 onetom, if you only want 64 words of stack then that's do-able. 08:54:37 which is what most forth chips seem to do... 08:54:46 what about ~16cells? 08:54:48 and for context switching you could only save the number of stack registers that you count (using the stack index register). that would be helpful. 08:55:05 tathi, yah. it's because you have to. 08:55:15 although i've never written a forth program that went more than 6 cells deep. 08:55:34 the return stack can b deeper of course and cached 2 slow ram some how 08:55:45 but experiments r required! 08:56:02 onetom, yah. you could keep the return stack in L1 cache. it's a lot slower than registers but you hardly ever access the return stack. 08:56:46 err, forth is all about factoring, wouldn't you be hitting the return stack constantly? 08:56:51 onetom, why do you abbreviate with stuff like u,r,2,b,4 ? it's hard to read that AOLspeak. 08:56:53 the programmer doesnt access the return stack so much programatically, but 08:57:08 tathi, no. you'd inline most things anyways. 08:57:14 it is accessed all the time intensively, coz the program runs... 08:57:20 at least all the forths i've used inline most things. 08:58:09 OrngeTide: ive told u that last day. it wont b so hard 2 decode some hrs of excercise, so dont worry about it 08:58:18 --- join: futhin (thin@h24-64-175-61.cg.shawcable.net) joined #forth 08:58:35 onetom, but it's not any faster to type. hitting 2 is a farther reach than typing to. :P 08:58:59 OrngeTide: but dont think in simulated 4th environments! simulation is always slower than the real stuff 08:59:36 i think the return stack is accessed most often for function CALLs. 08:59:38 OrngeTide: skip that topic, k? 4 a slow typer like me , thats shorter 09:00:02 but if you're inlining a lot of your tiny functions, which you will want to do to save on the overhead of doing a CALL then the return stack isn't a problem, right? 09:00:09 onetom, ehhee. okay. 09:00:23 OrngeTide: don't change onetom's typing style, i've come to like it :) 09:00:32 futhin, you guys are crazy:P 09:00:45 OrngeTide: but whats more important i did that type of writing so much that i couldnt get out of the habit of doin that easily 09:00:46 yep :P 09:01:18 onetom, back in the bbs days i used to do a lot of l33t sp33k typing. :) 09:01:46 OrngeTide: and the 3rd reason, i like it 2 ;) 09:01:51 the advantage is that if someone pops in here pretending to be onetom we will know if he doesn't type correctly:) 09:01:58 OrngeTide: 4th reason is under development ;) 09:02:44 OrngeTide: but I thought part of the point of a forth chip architecture was to make the overhead of a CALL as small as possible? 09:03:11 OrngeTide: k, let it b the 4th reason 4 keepin me style :) 09:03:16 since there's a return stack on the forthchip, all the "calls" to words are cheap 09:03:16 tathi, well it's slightly slower than a jump. 09:03:25 the return stack is used 09:03:27 1 cycle 09:03:52 tathi, which is pretty damn fast. but if you have a dictionary word that is only 4 instructions long then the overhead for call doesn't even save you memory. 09:04:28 ok 09:04:40 futhin: 1 think hasnt really been taken into account seriously 09:05:22 what do you mean by "save you memory" 09:05:35 there's 4 instruction prefetch cache 09:05:40 well a call takes more than 1 cycle. cuz you have to load a new instruction stream (and break your pipeline) and write out a word. 09:05:56 and writing on all the forth chips i've seen is not pipelined and rarely cached. 09:05:58 what pipeline?? this is forthchips we're talking about 09:06:12 futhin: mup21 has a pipeline. even though it's a weird way to represent one. 09:06:16 pipelines = lame 09:06:43 --- quit: ProteusGuy (Read error: 110 (Connection timed out)) 09:07:06 mup21 = dead 09:07:15 f21 & 25x = present and future 09:07:17 nah. they are good for trying to get your processor to do a memory io whenever possible. 09:07:43 well unless i'm mistaken the multiple instructions per word exists in f21 as well. 09:07:51 s/word/machine word 09:08:04 the forthchip has 4 instruction prefetch cache because 4 words is the common number of instructions that words have (if you code like chuck moore..) 09:09:11 basically any cpu that can load up multiple instructions and switch between fetching instructions or doing data io in any given cycle is pipelined. which mup21 did. and i assume f21 does as well. 09:09:19 oh crap. i'm late for work. 09:09:50 chuck just had a very simple and straight forward way to achive the same results as a pipeline, with fewer logic gates. 09:11:37 --- join: ProteusGuy (~proteusgu@24-197-147-197.charterga.net) joined #forth 09:11:51 you seem to be working with analogies.. 09:12:55 hm 09:20:05 --- quit: ProteusGuy (Remote closed the connection) 09:20:36 futhin: 4 a desktop computer, u need intensive multitaskin capability 09:21:05 futhin: so u cant relay on built in stacks unfortunately :( 09:21:16 built into the processor 09:24:17 relay? rely :) 09:32:33 P7 09:32:33 WIDTH 227 09:32:33 HEIGHT 149 09:32:33 DEPTH 3 09:32:33 MAXVAL 255 09:32:34 TUPLETYPE RGB 09:32:36 ENDHDR 09:32:46 its a PAM header 09:33:28 its the header of the most general 2/3d datarepresentation fileformat 09:33:43 pam - portable arbitrary map file format 09:34:36 its the most general 1 of the pbm pgm pnm ppm image-fileformat line 09:36:23 Portable Bit Map, Gray Map, Pix Map, aNy Map (see netpbm pkg) 09:36:59 oh, I thought it was a Pluggable Authentication Module. 09:37:17 i also thought some of u think of that :) 09:37:23 sot the question IS: 09:37:34 how can u write a loader 4 this 4mat? 09:37:44 in 4th ofcozz 09:38:16 it would b pretty easy 2 simply INCLUDE the file 09:38:28 if it would look like: 09:38:37 227 WIDTH ! 09:38:40 ... 09:38:43 Have a vocabulary for PAM, if it's that regular. 09:38:57 and ENDHDR could b a parsing word 09:39:01 * cleverdra nods. 09:39:41 any 4thish idea? 09:41:42 hmm, lets define those words as CR PARSE 09:41:50 what do u think? 09:43:11 Well, WIDTH and HEIGHT and TUPLETYPE seem to have data more interesting than a string. 09:43:57 certainly ithought of CR PARSE EVALUTE or CR PARSE NUMBER 09:45:34 but lets skip it 4 a while 09:45:58 what is the most effective techniqe: 09:46:33 define values like width and height as constants, variables or inline them? 09:47:10 --- join: Kitanin (~clark@h-209-91-66-234.gen.cadvision.com) joined #forth 09:49:02 if i wanna inline them 2 the image processing algorithm 09:49:34 i have 2 define the b4 compiling the program 09:50:00 so ENDHDR should do the compiling of the actual code 09:51:05 --- join: I440r (mark4@pool-63.52.218.43.ipls.grid.net) joined #forth 09:51:14 Hey I440r. 09:51:19 hi 09:51:31 pyromaniac !!! :) 09:52:29 I440r: i have a nice problem. can u read backlog? 09:52:53 err.... 09:53:23 P7 09:53:23 WIDTH 227 09:53:23 HEIGHT 149 09:53:23 DEPTH 3 09:53:23 MAXVAL 255 09:53:24 TUPLETYPE RGB 09:53:26 ENDHDR 09:53:31 given this text in a file 09:53:43 this describes an image 09:54:04 the actual imagedata comes after this header 09:54:09 --- quit: pyromaniac (Connection timed out) 09:54:11 how would u load it into memory? 09:55:18 erm. thats a database record ? 09:55:25 and you want to parse it ? 09:55:49 i would make WIDTH and HEIGHT parsing words and interpret the record 09:55:50 ? 09:56:26 what do u mean by db record? 09:56:36 its the beginning of an image file 09:56:41 oh 09:56:43 right 09:57:15 and i dont want 2 parse it :) the main goal is having it loaded in2 mem 09:58:07 certainly defining the mentioned word as parsing 1s 09:59:31 i dont understand your problem 09:59:56 eeyore: load this fuckin file! 10:00:04 nah, got it? :) 10:00:33 err. is the file binary ? 10:00:43 use a state machine 10:00:51 is it look binary 2 u? 10:01:08 r u sleepy or what? 10:01:34 cant u imagine an image file starting w the lines above? 10:01:55 well use a state machine 10:02:03 and after the word there comes the actual imgdata 10:02:15 we dont care bin or asc 4 now 10:02:38 state machine? why? 10:03:07 i didnt say parsing is not a good solution 10:03:07 state machines make parsing this easier 10:03:13 --- quit: cleverdra (Read error: 104 (Connection reset by peer)) 10:03:24 you HAVE to know what to expect next 10:03:28 what if someone puts 10:03:31 width 100 10:03:33 foo 200 10:03:33 i just said its not the goal, so u r free 2 choose anything else beside pasing 10:03:40 if u know something better 10:03:50 :) 10:03:52 i got it 10:04:04 finally ;* 10:04:06 you can do what i did in my assembler 10:04:06 or 10:04:10 you can make it do 10:04:13 100 width 10:04:17 200 height 10:04:22 eeeh, shit 10:04:24 and INTERPRET the file 10:04:34 my assembler makes it possible to have 10:04:36 i didnt say u can change the fileformat 10:04:36 width 100 10:04:39 its given 10:04:39 height 200 10:04:47 watch 10:04:56 variable ?a; 10:05:20 : a; ?a; @ r> ?a; ! >r ; 10:05:41 : width a; _width_ ! ; 10:05:49 : height a; _height_ ! ; 10:06:10 width doesnt store N in _width_ untl you execute the next word in the file! 10:06:20 u understand? 10:06:37 .... 10:06:41 (not yet :) 10:06:56 ok 10:07:27 i actually simplified the above a; word, it wont actually work liek that, 10:07:33 hmmm 10:07:34 ?a; is zero to begin with 10:07:37 watch 10:07:56 so u defer the execution of the next word or what.. 10:08:11 : a; ?a; @ r> ?a; ! ?dup 0> if >r then ; 10:08:39 : width a; _width_ ! ; 10:08:46 fetch from addr 0 ? 10:08:47 : height a; _height_ ; 10:08:54 no it wont fetch from address zero 10:09:05 it fetches from address ?a; 10:09:05 you said ?a; is zero 10:09:14 whichcontains a zero to begin with 10:09:15 ok 10:09:16 yeah 10:09:23 so 10:09:41 you EXECUTE the file and when WIDTH executes it runs a; 10:09:57 which pulls its return address off the stack and stores it in ?a; 10:10:09 why are you using ?dup? 10:10:10 when height executes it also executes ?a; 10:10:14 why ?dup 10:10:21 figure it out, its basic forth 10:10:32 but there's ALWAYS a number on the stack when ?dup is called 10:10:33 futhin: hold on. let him finish the explanation 10:10:40 so when height executes we have the return address BACK into width in the variable ?a; 10:10:41 so it doesn't need to check for a number 10:10:49 and we have the height value on the stack 10:10:58 erm 10:11:01 the width value on the stack 10:11:13 so when we execute WIDTH we store HEIGHT !!!! 10:11:50 it doesnt seems chainable 2 me 10:11:55 : P7 ?a; off ; 10:12:02 would be a good start 10:12:09 onetom you understand ? 10:12:26 it worx 4 2 consequent vars, but does it work in case of more? 10:12:36 yes 10:12:54 btw, it also strict about the order of lines 10:13:13 i440r: have you told chuck about this neat trick? :) 10:13:16 well this wouldnt be, but you could build in a test for out of order variables 10:13:25 futhin nope 10:13:53 i440r: it's a neat trick that avoids parsing, so i would think chuck would be VERY interested :) 10:13:56 k, lets skip the parsing problem 4 awhile 10:14:12 onetom its not parsing anything :) 10:14:18 how should i use the gained values in the program? 10:14:20 i440r: write up a short explanation of it and put it up on isforth.clss.net :) 10:14:44 i440r: today if you can.. (just a really short explanation.. ) :) 10:14:44 I440r: dont b dull, the 4th parses the file 10:14:48 futhin im not ready to do that yet, its part of my assembler and thats as yet not released 10:14:56 hm 10:15:16 onetom forth HAS to parse it, either you have to create a word to READ the file or you use existing words 10:15:24 onetom: nahhh, forth interprets the file :P 10:15:28 this saves you from writing any file reading words 10:16:23 futhin: parse+exec=interpret 10:16:26 i440r: why do you call them a; and ?a; ?? 10:16:46 futhin: probably because they r used in asm 10:16:57 futhin: a is 4 address iguess 10:17:38 what would be a good name for a; and ?a; if they are used outside of the assembler? 10:18:10 putoff puttoffvariable 10:18:11 heh 10:18:14 ugh 10:22:45 futhin its a refinement of a technique used in PASM (fpc's assembler) 10:23:02 a; is like executed at end of each asm instruction 10:23:13 a; = end of instruction 10:24:05 onetom you CANNOT read this file in without parsing it in SOME way. why write "special" words to parse this file type, just use forth 10:25:27 gotta go 2 the shop but bb soon 10:26:10 I440r: if u use the 4th word PARSE u still use 4th ;p 10:26:19 i thought of sg like: 10:26:48 : WIDTH CR PARSE EVALUATE WIDTH ! ; 10:27:37 i would use : width parse here number not if abort" blah" then ; 10:27:40 or something like that 10:27:46 erm forgot to store it :P 10:29:21 yeah, yeah thats faster, i thought it too 10:29:46 but i have 2 admit, i never used number :D 10:30:26 but, now, lets see the next question 10:30:56 how should i store & use this info during the program 10:31:05 there r 3 possible solutions 10:31:15 define them as constanst 10:31:19 define them as variables 10:31:29 inline them as literals 10:31:46 which 1 is the fastest, eg in isforth? 10:32:11 inlining also turns the problem upside down 10:32:31 coz u cant have the application already compiled 10:32:46 at the point of reading image dimensions, so 10:33:01 i440r: yeah, i was wondering if you told the a; trick to chuck moore, that chuck moore might be able to see a whole new world open up and would create a whole new spinoff of forth ;) 10:33:13 eg, ENDHDR should do the loading of the main program 10:34:38 * onetom -> shop 10:41:08 futhin you tell him :) 10:41:09 brb 10:44:04 i440r: is it an original idea or not? would he already know of it? 10:44:50 back 10:57:55 --- join: cleverdra (julianf@0-2pool111-57.nas2.florence1.sc.us.da.qwest.net) joined #forth 11:00:20 : WIDTH s" width" ; 11:00:58 : ENDHDR ( s" width" is on stack and so is the number next to width) 11:01:35 er 11:01:35 no 11:01:44 variable widthvar 11:01:52 : width widthvar ; 11:02:01 : endhdr swap ! ; 11:03:34 * Kitanin smacks himself in the head. 11:04:32 what? 11:04:47 Five days staring at this code, refusing to even _think_ about FILE-SIZE returning a 2-cell result... 11:05:09 * Kitanin misses blocks. :-) 11:05:27 what 4th r u using? 11:05:54 gforth. 11:06:21 futhin: ah, got it. sure, thats the simplest solution, tho it requires a specific order of the params, what is not granted unfortunately, ithink 11:06:26 Kitanin: 0.5.0? 11:06:35 kitanin: what code? 11:06:41 Ayep. 11:06:44 Kitanin: coz that should support blocks 11:07:41 I know it does. The problem is, the ``problem'' doesn't. :-) 11:07:51 onetom: sure it would work, widthvar address won't get placed on stack if there isn't a WIDTH in the parameters.. all you have to do is code the ENDHDR word to test how many things are on the stack, and then repeat SWAP ! with all of them, or something similar anyways 11:08:59 wow :) 11:09:03 so long as the parameters are ALWAYS like: WIDTH 222 or XXX 232 11:09:19 you could probably take into account for the exceptions to the rule 11:09:20 definetly not a bad ida 11:09:21 idea 11:09:31 :) 11:09:55 but 4 string parameters we can avoid parsing ithink 11:10:19 cant 11:10:29 eh, i always miss the T 11:14:16 * onetom rediscovers the netpbm problem domain 11:32:19 --- join: Kiara (~clark@h-209-91-66-234.gen.cadvision.com) joined #forth 11:32:19 --- quit: Kitanin (Read error: 104 (Connection reset by peer)) 11:32:38 Yep, the code works much better when I treat FILE-SIZE as returning a ud like documented... 11:32:44 --- nick: Kiara -> Kitanin 11:32:56 kitanin: how long have you been familiar with forth? 11:33:46 Fifteenish years. I've just never actually used files with FORTH before. It's been all blocks. :-) 11:34:28 :) 11:35:18 --- join: pyromaniac (~pyromania@dialup-111.159.220.203.acc01-high-pen.comindico.com.au) joined #forth 11:35:20 Kitanin: and what do u think of those previous 3 questions? 11:35:52 Which 3? 11:36:13