00:00:00 --- log: started forth/12.02.26 00:50:17 segher marcel hendrix writes optimizers that do BETTER than most c optimizers 00:50:27 so ya 02:09:32 on some code, yes 02:10:30 on real-life good code (good C code and good Forth code, not 1-1 translations!) it's a wash 02:16:52 --- join: bzimmerly (~bzimmerly@96-35-182-216.dhcp.stls.mo.charter.com) joined #forth 02:16:53 --- mode: ChanServ set +v bzimmerly 02:17:57 --- quit: bzimmerly (Client Quit) 04:58:37 --- join: tathi (~josh@dsl-216-227-97-239.fairpoint.net) joined #forth 04:58:37 --- mode: ChanServ set +v tathi 05:01:09 But in a practical sense, there aren't any free Forth compilers of even reasonable quality that I know of. 06:05:50 --- join: Kumul (~Kumul@67.224.230.59) joined #forth 06:05:50 --- mode: ChanServ set +v Kumul 06:19:03 there is some specialised stuff 06:19:17 also, i want to change that :-) 06:44:35 Yeah, that would be nice. 07:22:48 i have a compiler that generates native code for two quite different archs now (u'nsp and xs1) 07:23:19 and working on optimiser 07:24:25 it's really quite easy, compared to e.g. C or fortran or whatever 07:24:39 esp. register alloc is much much easier 07:25:05 it is always optimal (or close to optimal) to spill the oldest stuff when you run out of registers 07:25:57 the hardest part is tuning the inliner, heh 09:46:14 tahti ima smack u lol ISFORTH is free! and almost good ish kinda... . . 09:46:48 segher i think i need to has a chat with u regarding your native compilers 09:47:46 i think you need to learn how to spell :-) 09:50:49 that too 09:51:11 plus i can type 35 wrong words per minute 09:51:52 i just created a direct threaded AVR forth. wrote the assembler with isforth and then a kinda kludgy target compiler. (that needs work) 09:52:09 what i want to do is release that as free (to adafruit if you know who they are) 09:52:21 and have a native compiler version as shareware 09:53:04 never did a native compiler before :) 09:56:10 if you mostly do STC it is easy 09:56:27 maybe inline a few hand-picked things 09:56:28 ALL of my forths are direct threaded 09:57:00 most of mine are native or ITC 09:57:07 i can think in direct threaded. processor stack = parameter stack. 09:57:17 ITC is faster than DTC on big machines, by quite a big margin 09:57:20 i see no need for ITC, not even on a harvard architecutre device 09:57:43 segher that depends on weather or not code/data interleaving is a performance hit 09:57:53 not only that 09:58:18 your branch target predictor is the main thing 09:58:55 well. the person writing the code would have a good idea how to predict most branches 09:59:02 maybe 09:59:19 every 6th insn is a branch, eh 09:59:41 less if you don't inline enough :-) 10:00:10 call and returns are branches too 10:01:11 yes any change of program flow is a branch 10:02:27 and ITC is much easier to predict the branches, because there are far fewer 10:02:57 only one per word instead of one every time the word is called 12:57:21 --- join: MayDaniel (~MayDaniel@unaffiliated/maydaniel) joined #forth 12:57:21 --- mode: ChanServ set +v MayDaniel 13:00:29 --- quit: MayDaniel () 14:06:33 --- join: _spt_ (~in@unaffiliated/-spt-/x-5624824) joined #forth 14:06:33 --- mode: ChanServ set +v _spt_ 14:09:02 --- part: _spt_ left #forth 14:51:20 --- quit: ttmrichter (Quit: Leaving) 15:03:17 since ASau doesnt believe forth can compare to C maybe it can compare and be a better alternative to a scripting language? (pardon my ignorance) 15:04:01 What do you call "a scripting language"? 15:05:07 the original definition (per TCL) 15:06:04 If you compare Forth to TCL, Forth is just incapable. 15:06:24 At most you have prototypes and conceptual things. 15:06:39 In TCL you have readily available software. 15:07:36 Note that in Forth first problem you're going to solve is string processing library. 15:07:44 Isn't it wonderful? 15:22:53 readily available software? what do you mean? packages? 15:23:26 Packages too. 15:25:17 that would be easily solvable, so long as there is a community willing to make them 15:55:44 Kumul: the thing is that a lot of (most of?) the people who use Forth seem to subscribe to the theory that premature generalization is the root of all evil. :) 16:12:35 you use Forth because you want to have a tight and tiny system 16:13:39 if you do not see why that is a Very Good Thing, even on 6GHz 64GB 128core machines, all hope for yyou is lost 16:17:47 forth does string processing better than any other language ive ever seen 16:19:23 i think the only reason asau hangs out in here is to scare away anyone who might have an interest in forth 17:49:39 it is interesting to note that all the _modern_ C string libraries use the same representation as Forth (addr,len pairs) 18:09:23 --- join: adasfdasf (poxa@200.185.225.95) joined #forth 18:09:23 --- mode: ChanServ set +v adasfdasf 18:09:24 --- join: amca (~amca@CPE-121-208-82-128.cqzr1.cha.bigpond.net.au) joined #forth 18:09:24 --- mode: ChanServ set +v amca 18:10:06 hi, can forth be used to do a web framework? 18:10:21 yes 18:10:23 it has already 18:10:54 where? 18:11:12 thers web servers written in forth. does that count? 18:11:55 i also think forthfreak had something to say about doing emedded forth in html 18:11:58 i forget 18:12:17 i know ive heared of that somewhere other than my own immagination :) 18:12:33 okay, once forth can be implemented in any language (fully ANS-compilant or as simple stack machine), can someone create a FORTH-like-web-framework using PHP? 18:13:05 just to reuse the lamp stack, what do you think about? 18:13:21 yea that would be doable 18:13:32 but it wouldnt be forth 18:13:41 i dont think forth written in C is forth 18:13:57 and ans compliance is something i strive NOT to achieve 18:13:59 lol 18:14:03 heh, real forth is done in forth 18:14:14 exactly 18:14:57 I currenctly use C to do computer vision software, I use OpenCV 18:15:42 I looking for new models of abstraction, new manner to write code 18:16:02 that sounds pretty interested! 18:16:15 you can do a reasonably efficient Forth interpreter in modern javascript 18:16:25 (it wasn't possible two years ago) 18:16:36 thers bashforth thers no reason you CANT develop forth in anything other than forth and asm 18:16:37 needs typed arrays eh 18:16:44 problem is, thers no reason you should 18:17:02 also, I'm looking for a way to run away from any OS and his complex issues, for my apps 18:17:02 sure there is. i cannot write javascript :-) 18:17:23 adasfdasf, this an embedded system? 18:17:42 can forth be used to deal with images? image formats? cams? or need I recreate it from strach on forth? 18:17:57 forth can do anything the coder can make it do 18:18:02 not today, but it will be 18:18:09 it can also probably do it better than ANYTHING ELSE 18:18:26 chuck more says forth is *T*H*E* software engineering solution 18:18:28 there are not very many high-quality libraries available; in the open source, close to none 18:18:29 i agree 18:18:49 most vision and image processing algorithms are VERY closed source 18:18:56 yes 18:19:04 if adasfdasf does image processing he has algorithms 18:19:21 hey, i write my C as i write Forth these days, and it is soooooo much better for it 18:19:27 ive done a small amount of image sensor code tho not in forth 18:19:50 would be nice it! FORTH as a programming model, even in C 18:20:08 no. he means the methods he learned as a forth programmer 18:20:14 help him produce better c 18:20:25 i.e. instead of if/and/but loops nested to the umpteenth level 18:20:31 you write VERY VERY SMALL primatives 18:20:39 and use those as building blocks which yo use as building blocks 18:20:59 when you are programming in Forth, you are making a solution for the problem at hand; in more fashionable languages, you just stack abstraction on top of abstraction until the whole thing collapses (and the project gets cancelled) 18:21:00 is it like a dict? 18:21:18 adasfdasf, thats exactly what it is 18:21:32 there is technally no such thing as a forth program 18:21:32 i440r: yes, i write very small functions. but there is more than that 18:21:39 what you do is you EXTEND forth with your application 18:21:44 thrers FORTH and thats it 18:21:46 i do not "stub out" functionality, ever 18:21:52 you just defined forth to solve your particular problem 18:21:54 implement bottom-up 18:22:14 also, TEST EVERYTHING AS SOON AS YOU WRITE IT 18:22:16 design is best done top down. implementation is best done bottom up 18:22:24 no forward references if they can be avoided 18:22:25 EVERY SINGLE LINE, pretty much at that granularity 18:22:32 yup 18:22:37 soory for shouting, but this is soooooo important :-) 18:22:50 heh 18:22:57 its a golden rule 18:23:04 : foo small code here ; 18:23:06 1 2 3 foo 18:23:08 ack that wrong 18:23:10 i don't know if i call my design top-down. it's usually very hand-wavy 18:23:15 : foo different code here ; 18:23:17 1 2 3 foo 18:23:19 yay it works 18:23:33 and never ever do unit tests 18:23:39 unit tests are busy-work 18:23:43 at every stage of development everything you have already developed has already been tested 18:23:45 I do not have good experiences with bottom-down design, but it was using languages like Ruby in web applications, I did Test-Driven and I've found it very poor and slow. 18:24:04 if your components are fragile, you did something Wrong 18:24:38 "bottom-down design"? hehe 18:24:45 arse on the chair? :-) 18:24:51 ops, bottom-up 18:25:05 bottom-up _implementation_ 18:25:09 yes 18:25:30 there isn't design top-down 18:25:49 oh there sort of is. i usually have a plan of attack 18:26:07 it doesn't always happen that way in the end though :-) 18:26:50 why isn't FORTH so popular? 18:26:56 remember: any piece of code that hasn't been tested _has not been finished yet_, "does not exist" 18:26:59 number of reasons 18:27:07 reverse polish scares nubs away 18:27:11 thers also another reason 18:27:14 it isn't reverse polish 18:27:18 lets say you buy msc version 1.0 18:27:23 and i buy ms forth version 1.0 18:27:32 we both did this sometime in the 80s 18:27:38 you report some bugs 18:27:42 10 years later they are still there 18:27:42 heh 18:27:47 not enough people reported them 18:27:48 k 18:28:03 ms fixes some other things tho and you buy the upgrade 18:28:09 mathematicians work with symbols and obscure ones, they dont like words (perhaps?) 18:28:11 me.? i see some bugs, fix them and keep going 18:28:26 you request some new features. ms says yea lots of ppl asked for that and implements them 18:28:32 then they charge u for the upgrade 18:28:45 me? i see a deficiency or a feature i want. i code it and add it to the system 18:28:48 scan forward 20 years 18:29:04 you have close to 500k of money invested in your msc compiler over the years 18:29:14 i think one of the reasons people nowadays do not use Forth as much as they should, is that Forth was a very popular language on the micros of the 80s 18:29:15 i have the $50 i spent to get version 1.0 18:29:28 no way microsoft is going to market something like that 18:29:46 and those implementations were always quite spartan, for obvious reasons 18:29:54 another reason is because C was invented at bell labs and bell was a HUGE corporation and PAID people to develop operating systems and libs using it 18:30:13 forth is used 18:30:25 and those small cheap machines _themselves_ were quite spartan, making it worse 18:30:26 but today every programmer must to know OO 18:30:38 until about 5 years ago all the satellite tracking systems used by direct tv were all coded in forth 18:30:41 the entire system 18:30:49 the entire fedex tracking system is forth 18:30:55 just to reuse and to work with the system 18:30:56 so now people have a distaste for Forth, remembering the Forths they used then; analogous to BASIC 18:31:07 object obfuscation usually destroys more projects than it helps 18:31:23 it is not like modern BASIC is worse than PHP really, quite the opposite 18:31:42 there are no 3846587364598698 gigs of libraries written for forth 18:32:00 if you want a tcp/ip stack you create it 18:32:07 with c you just pay someone for theirs 18:32:13 or download a free lib 18:32:23 you dont have to THINK to code c. forth makes u think 18:32:43 you NEED TO think to code C. a lot. and most people don't, alas 18:32:48 thers no "learn forth in 24 hours" 18:32:55 c discourges you from thinking 18:33:05 that's C++ 18:33:06 hey dont think about how this works. just USE it 18:33:14 so If I need to deal with IP cam, to deal with JPEG format, need I create these libraries (or words) from scratch on forth? 18:33:23 any langyage allows you to shoot yourself in the foot. 18:33:26 oh sure, yes, it's an abstraction\ 18:33:27 thats insane, a programming language cannot discourage someone not to think 18:33:30 C and its ilk require you to do so 18:33:32 repeatedly 18:33:36 maybe Chuck Moore doesn't believes in software reuse 18:33:47 "Developing embedded applications in C is like opening a can with a rock" 18:33:57 adasfdasf, sort of 18:34:13 what he believes is that code you write for one system is NOT portable to a different system 18:34:23 ada: there is software reuse and software reuse. i reuse most of my Forth code (and C code). i usually rewrite parts of it though 18:34:26 you can mash that square peg into the new round hole but it wont fit exactly 18:34:33 and development time in forth is NEGLIGABLE 18:35:14 the time it takes to rewrite an application from scratch is a complete and utter NON ISSUE in forth 18:35:27 it always is 18:35:37 its a non issue in forth. its a major issue in c 18:35:44 but in c where they claim they have portable code 18:35:44 nope 18:36:00 code that compiels and runs on windiws. fbsd, linux. wherever they DONT have portable code 18:36:01 if you want to use big libraries in C, you have that problem, sure 18:36:16 yes, I know 18:36:19 I heve 18:36:21 what they have is 34658639 different versions of the same thing interleaved intoa clusterfuck of heavilly obfuscated conditional compiation 18:36:22 I *have 18:36:34 that clutters up the entire source tree 18:37:09 lots of C code is like that, or bad in other ways, yes. it does not have to be though, it's not the language's fault 18:37:10 i wont suppory conditional compilation in isforth 18:37:11 not ever 18:37:15 I do software reuse in my own software, and every time I need a monkey I got the jungle together. 18:37:39 segher no it IS the languages faut. the language is just SOOOOOOOOOOOOOO easy to fuck over in that way 18:37:40 in fact, the C language *encourages* you to work close to the metal, like Forth 18:37:46 becayse thats the way everyone fucks it over 18:37:53 no. it doesnt. 18:37:56 it allows you 18:37:57 and it encourages you to tread carefully 18:38:08 the c mentality is to disassociate yourself from everyt8ing 18:38:10 that's why there are traps all over the place eh 18:38:19 C is like guitar 18:38:21 I440r: why? 18:38:28 the guitar is probably one of the easiest instruments to learn 18:38:32 C is like electrified guitar, heh 18:38:36 every fucking moron AND his autistic brother all code C 18:38:44 yes 18:38:46 jajaja 18:38:50 guitar is also probably one of the single most difficult instruments in the world to master 18:38:50 and they are not rogrammers 18:38:54 ditto C 18:38:57 exactly 18:39:18 c allows and enables you to write very beautiful code 18:39:33 except the pointer and memory management, I can see C in javascript, actionscript, PHP, even Perl, Ruby. 18:39:33 it encourages you to write cluster fuck obfuscated source puke 18:39:41 wtih if/and/but loops nested to the umpteenth level 18:39:46 but it is hardly the language's fault that it is so popular. that's mostly history, and partly the ecosystem surrounding it 18:39:49 c is the single worst language ever devised by man 18:39:57 it IS the languages fault 18:40:13 because it makes it VERY VERY easy to create BAD CODE 18:40:22 so does Forth 18:40:22 i consider for example the sources to the linux kernel BAD CODE 18:40:26 forth allows 18:40:37 it very VERY strongly encourages good code 18:40:43 proper factoring 18:40:44 --- join: malyn_ (~malyn@unaffiliated/malyn) joined #forth 18:40:44 --- mode: ChanServ set +v malyn_ 18:40:48 I don't think so. C is used today as different form than was designed. 18:40:48 bottom up development 18:40:58 although, even a modestly complicated thing in Forth, written by someone who fucks about, will never ever xo anything 18:41:05 --- quit: malyn (Quit: Disconnecting from stoned server.) 18:41:36 why is linux kernel BAD CODE? sorry but I dont know FORTH code yet to match the coding forms. 18:41:38 segher im not saying you cant write complex code in forth thats difficult to read 18:41:41 --- nick: malyn_ -> malyn 18:41:47 im saying C pretty much FORCES you to do so 18:41:50 oh i know what you are saying, and we agree 18:41:57 forth encourges youy very strongly to NOT do so 18:42:01 we just don't agree on your conclusion :-) 18:42:13 what about xv6 ? 18:42:30 this is because segher still uses C 18:42:40 well. have you worked as a consultant software engineer for over 25 years during which time you have waded through some horrendously BAD c code? 18:42:42 i have no idea what xv6 is 18:42:50 sure 18:43:02 well, 18 years or so 18:43:10 http://pdos.csail.mit.edu/6.828/2011/xv6.html 18:43:53 segher an example of some not so good forth code is the isforth memory manager 18:44:05 I440r: do you work today exclusively with FORTH? 18:44:06 thats quite complex and makes use of some hax0ry methods 18:44:18 no i work exclusivly with it on my OWN projects :) 18:44:32 i do c/asm when i work a contract 18:44:41 currently not working 18:44:47 interview tomorrow for a contract tho 18:44:52 good luck 18:45:29 dunno if i want the job. ill take it if the AVR job doesnt materialize :) 18:45:37 I440r: do you need to deal with C++ code, like in windows development? 18:45:50 no i COULD tho 18:46:09 segher? 18:46:21 C++? i stay away from it like the plague 18:46:33 but I touch in it. 18:46:36 haven't written it for 15 years 18:46:51 i recently ported the BOOST libraries date/time code into a non posix compliant embedded os 18:47:11 why on earth use C++ on embedded, oh the humanity! 18:47:19 not my call 18:47:27 why use C 18:47:38 because they can hire 284593 morons to work on the produt 18:47:53 yah 18:47:55 now I'm working with Kinect and Computer Vision, getting hand gestures from it, all the libraries from Microsoft are C++, I have no escape. But I wrote some "C with Classes", do my job like C using the resources from libraries 18:48:22 heh ladyada was the one that offered the prize for the first kinect driver 18:48:27 sure you have an escape. the open source drivers are not C++ 18:48:38 and marcan got it :-) 18:48:48 ms told her... err yuo cant do that its ours and you have to keep your hands off 18:49:03 the prize was 2k and whem ms got pissy about it 18:49:10 she said erm ok. ill bump it to 3k 18:49:12 LOL 18:50:32 the main library that I use to do my Computer Vision development in Windows was created in C by Intel, now it's going to be ported to pure C++. The already done algorithms now are 10x more slow. I'm using an outdated version of this library. 18:50:33 i have an atmega32u4 breakout board here form adafruit 18:50:38 i developed an avr forht for it 18:50:55 adasfdasf, intel does not develop image sensor code! 18:51:01 are yu sure that isnt visionmos code? 18:51:31 intel used to have an image sensor division. they dont any more 18:51:32 it's OpenCV, or IntelCV 18:51:40 now out of Intel, opensource 18:51:44 hrm i dont think thats visionmos then 18:52:24 visionmos is 4 guys that used to work for motorolas image sensor division, one of them RAN the division 18:52:30 then they went to work for themselves 18:52:47 4 guys but they have customers like magnachip who are the worlds third largest producer of image sensor devices 18:53:16 so, what's the best way to learn forth? 18:53:25 bu using it 18:53:36 i would give a shameless plug here for isforth :))) 18:53:41 assuming u have an x86 linux box 18:53:47 jonesforth is <3 18:53:59 I got gforth 18:54:01 also, starting forth 18:54:03 jonesforth is for 68k? 18:54:09 gforth is not forth 18:54:14 and its ans. that makes it doubly not forth 18:54:14 write something that you would do in a bash script, or Perl script, or small one-file C code, as Forth. 18:54:39 and write it _in a Forth system_,not in an editor 18:55:01 copy-pasta it into the editor when it works, of course 18:55:03 why isn't gforth forth? 18:55:15 because its ans and because its written in c 18:55:22 gforth is not a very developer-friendly system 18:55:23 so all these forths are done in forth? 18:55:25 so is jonesforth i believe 18:55:38 isforth is done in asm and forth 18:55:47 jonesforth as well 18:56:00 jones is asm/forth developed? 18:56:01 not c? 18:56:05 yepp 18:56:05 i stand corrected then :( 18:56:07 :) 18:56:26 is jonesforth ppc or 68k or something? 18:56:27 i ferget 18:56:33 what's the name of that small forth not-ANS written in C? 18:57:00 there are many so called forths written in c 18:57:15 I was reading gforth manuals and Thinking Forth 18:57:43 https://github.com/chengchangwu/jonesforth/blob/master/jonesforth.S 18:58:31 adasfdasf, if you have a linux 86 box with svn 18:58:51 svn co http://24.183.5.217/svn/isforth/trunk isforth 18:58:54 username is guest 18:59:00 no pass required 18:59:09 thats a more up to date version than the one at www.isforth.com 18:59:43 brb, reboot 18:59:55 --- quit: I440r (Remote host closed the connection) 19:01:48 what do you think about http://www.murphywong.net/hello/simple.htm ? 19:02:02 --- join: I440r (~zhiming@24-183-5-217.dhcp.fdul.wi.charter.com) joined #forth 19:02:02 --- mode: ChanServ set +v I440r 19:04:03 so is jonesforth not ANS Forth ? 19:05:37 what I need to run this .S file ? what is it? 19:05:55 .S is asm source 19:06:49 so forth done in asm? 19:07:01 isforths kernel is done in asm too 19:08:17 jonesforths primatives are not as efficient as the isforth ones 19:08:25 it is direct threaded tho yes? 19:08:30 why need I ANS Forth ? 19:08:52 i have nothing good to say about ans forth 19:09:13 this jonesforth however adasfdasf would be a VERY good place for you to start 19:09:19 the source code isnt just soruce code 19:09:31 there is a tutorial inside 19:09:32 its a very very in depth explanation as to how everything in there workds 19:09:35 yes 19:09:40 a very good one 19:09:59 so, the last question about forth 19:10:07 for while 19:10:17 dont stop asking :) 19:10:22 why don't use text editor? 19:10:34 so, I tried the command line interpreter of gforth 19:10:37 all nice 19:11:20 but when I wants to save the software, I need to rewrote it into small file, to reload again. or forth have a better way to save the dict? 19:12:15 what os is jonesforth for? 19:12:40 linux, I think 19:13:01 --- quit: amca (Quit: Farewell) 19:13:04 this link is for Linux i386 19:13:07 gasm 19:15:41 isforthuses nasm 19:16:10 i think isforths soruces are easier to read but i stillsay that jonesforth is good. good tutorial! 19:16:40 this is not working: ./jonesforth: Killed. 19:17:31 pretty tutorial inside code 19:22:59 this jonesforth doesn't works in new linux systems 19:33:20 isforth will 19:33:27 if u disable heap randomization 19:33:38 somthign i need to fix later 19:35:17 ada: my point is that you should test everything immediately, i.e. write it on the command line. when it works as it should, you should of course copy it to your editor. if you write things in your editor, wou will write too much without testing 19:35:42 do you do gui programming in forth? 19:35:53 no 19:35:55 its all console 19:36:03 i need to write an xlib interface 19:40:36 this tutorial inside source is about create a forth, not about forth language itself 19:42:42 there isn't command line in this jonesforth, duh! 19:42:45 it shows u how each aspect of forth works 19:42:57 there is in isforth :) 19:44:37 so, "until you understand how FORTH is written, you'll have only a very superficial understanding of how to use it" - is it true? 19:45:25 yes 19:45:33 i say that is true of all languages 19:45:42 and thats another thing c discourges u to know 19:46:03 u will learn more about the internal workings of forth in 1 week 19:46:09 than 25 years of c teaches u about c 19:47:32 of all languages? so I want to learn lisp, go to read the source code of elisp? 19:48:29 thing is. read isforth soruces and understand isforth 19:48:36 read gcc sources and be lost 19:51:08 is this repo official ? http://24.183.5.217/svn/isforth/trunk 19:52:17 its on a server 2 feet to my left :)( 19:53:17 what it differs from isforth.com ? 19:53:45