00:00:00 --- log: started forth/02.04.19 00:00:40 lol 00:00:48 you know your ANSI comment about people driving out of their lane is an excellent analogy 00:01:00 although you meant to type "antsy" 00:01:43 erm 00:01:45 \i know 00:01:45 it doesn't say what type of car you have to drive, or even that you have to stay within the lines, but that you can expect trouble with other cars if you don't. 00:01:49 i always type it that way 00:01:59 and i know the correct spellin too heh 00:02:26 no, with ANSI Forth, I mean ... it's a good analogy 00:02:46 actually -with me it is heh 00:03:42 I don't understand why it is that people read it, and think it somehow dictates how to build their forth 00:03:55 I watched it happen again today 00:04:10 with herk 00:04:59 heh 00:05:08 in fact, he was wondering why they didn't do it the way f83 did and just dictate the details of how the words worked 00:05:29 instead of being all vague and obscure 00:06:27 he kind of bounced between the rails on that one 00:06:27 alot of what was done in f83 is fucking awesome 00:06:36 yes indded 00:06:48 but the standard was terribly written 00:07:00 yes 00:07:06 the standard was. 00:07:12 but then. ALL standards are 00:07:19 --- join: Soap` (flop@210-54-74-221.dialup.xtra.co.nz) joined #forth 00:07:30 I think ANS is much better in that regard, but you have to think about the vague terminology 00:07:42 yes, still not perfect 00:07:51 it never will be 00:08:22 i realy relay think the sole purpose of the std is to make forth more acceptable to the c crowd 00:08:39 do you think that the committee should use a system similar to RFCs for various libraries now? 00:09:10 I myself notice that it's easier to port from one forth to another now 00:09:33 rfc's are a good idea 00:09:43 I440r: an RFC type system would be similar to the "publication standard" that Moore was after 00:10:42 there would be a bazzilion little APIs to stuff like sockets, regex, timers, semaphores, serial ports, etc etc 00:10:57 graphics, windowing widgets 00:11:02 sound 00:11:41 exactly whats missing from forth 00:11:49 things i plan to add for isforth 00:12:08 then, on whatever platform you were working on, you try to write the commonly used APIs 00:12:22 if yoiu can't do it, no biggie 00:12:44 but if you do, you know your favorite whiz-bang app will be able to use it 00:14:03 you'd probably end up with vocabularies named like "SOCKETS-1.2" 00:14:21 and "SOCKETS-0.9" 00:15:28 SOCKETS-1.2 would probably load SOCKETS-0.9 ... alias many of the words into itself, and then redefine a couple for added functionality 00:15:39 heh 00:15:44 does that seem appropriate? 00:15:51 not realy 00:16:19 i would hope sockets 0.9 could be written to account for all the shit 1.2 needs 00:16:31 well, it's wasteful in name space, rather efficient in code space ... and seems like the way libraries would probably evolve 00:16:52 name space waste is acceptable 00:16:57 turnkey dumps headers 00:16:59 well, ok, an example then ... 00:17:59 in the 80s, a 9,600 or 14,400 baud limitation on the serial line was not uncommon ... and drivers were written with that in mind 00:18:24 later, 119,200 became the standard limit, but many cards went higher 00:18:53 and then came a slew of 4 and 8 port cards that all shared one interrupt 00:19:00 rite 00:19:20 noway could a writer in the 80s know of a proper API to handle all that universally 00:19:45 fossil! 00:19:49 although such apis are pretty common now 00:20:02 yep, fossil evolved as the standard 00:20:05 fido opus seadog sandard interface layer!!! heh 00:20:40 it still didn't handle multiport cards, but it was an excellent start 00:20:46 rite 00:21:26 so you can see how 2.0 of an API might be substantially diffferent from 1.4, but still need the same code? 00:22:29 well one just adds new code to the old source in forth 00:22:40 all the old code is still there but now with new words 00:23:02 ah! but if I loand 1.4 ... does it or does it not have the added functionality? 00:23:20 if I loaded 1.4 00:23:52 lol 00:23:59 and don't tell me to go read the code ... that doesn't make sense with libraries 00:24:12 if it doesnt and you need it - the compile will abort with an error message that tells all heh 00:24:22 or 00:24:31 are you talking about releasing eh object 00:24:34 not the source 00:24:51 or the words won't act like I'm expecting them to, causing run-time stack errors 00:25:13 well, I was thinking libraries distrod in source form 00:25:24 rite 00:25:40 the source file could be 00:25:42 but the point of a library is that you can drop them in, read a few pages of docs, and start using them 00:25:59 some-library-1.04-with-gizmoze.f 00:26:10 not have to go mucking about in the code or having to understand how the whole thing works 00:26:57 I, for one, use IP stacks all the time with sockets, but I do *NOT* want to understand how IP stacks work 00:27:19 er. neitehr do i 00:27:24 thats operating system shit 00:27:36 forth shoud be able to interface to the operating system 00:27:39 not replace it 00:27:51 yes, but it used to be that OSes didn't have IP stacks 00:27:58 if you wanted one, you had to write it 00:28:06 now they're standard 00:28:16 and the new thing is RPC via XML 00:28:18 thank god 00:28:29 very useful thing, RPC 00:28:36 a real headache to implement 00:29:21 rpc ? 00:29:41 VOCABULARY MYAPP ONLY FORTH ALSO RPC ALSO SOCKETS ALSO MYAPP DEFINITIONS 00:29:53 Remote Proceedure Call 00:31:14 hmmm ... 00:31:50 aha 00:31:52 as in 00:31:56 you over there 00:31:59 executing code over here 00:32:09 client side as opposed to server side 00:32:49 ['] my-tuple-handler s" SELECT name,address,phone FROM employees WHERE workday != Sunday" SQLDatabase RPC throw 00:33:11 --- join: davidw (~davidw@ppp-178-10.25-151.libero.it) joined #forth 00:33:54 what APPEARS to be a local procedure call is actually transported and executed on another machine or process 00:34:35 im not sure i like that idea myself. i think client side shit is evil myself 00:34:45 another example ... gettimeofday might fetch from the local hardware clock, the local software clock, the clock on my intranet server, or from tock.navy.mil 00:35:37 thats not executing gettimeofday on tock.navy.mil tho 00:36:10 sure it is ... although there might be another handler in there somewhere 00:36:21 what was your issue with client-side? 00:36:28 it would be alot more effcient to just do 00:36:33 "hey what time is it" 00:36:40 do a request. then get a response 00:36:43 instead of saying 00:36:50 hey you urn "xxx" for me 00:37:04 then it rns xxx and sends you the output 00:37:05 ugh 00:37:22 yes, that's basically what RPC is 00:37:47 i dont fukcing understand why people have to fucking tie things up into these big ass gordian knots 00:38:00 essentially, you "marshal" all the parameters into a data packet, send it somewhere, and expect an answering data packet in return (or not) 00:38:00 s" is a prime example of a fucked up way of doin things 00:38:06 s" xxx yyy zzz" evaluate 00:38:09 fucked up 00:38:13 : foo 00:38:17 xxx yyy zzz ; 00:38:24 THIS can be turnkeyd 00:38:28 heh 00:38:52 ok, I want a centralized database ... how should I do that? 00:38:54 there is absolutely no priactical application for the word s" 00:39:02 what so ever 00:39:08 huh? 00:39:10 ther are BETTER ways to do the same thing 00:39:13 what's that? 00:39:20 there is absolutely no priactical application for the word s" 00:39:35 heh, I read what you wrote, just didn't understand it 00:39:40 do you have " ? 00:40:43 actually no. im still trying to decide what to define 00:40:57 i have ," and my ,' invention heh 00:41:01 erm i might have " 00:41:04 a forth without some type of string literal seems rather crippled to me 00:41:12 i actually think some of these words are mis-named 00:41:31 ok, do you want 'adr cnt' returned or adr of counted string? 00:41:46 ," 00:41:54 compiles a counted string 00:41:58 create foo 00:42:01 ," xxx" 00:42:06 can I do ... 00:42:06 foo returns the address 00:42:08 i could do 00:42:15 create foo 00:42:15 : foo ," blah" ; 00:42:16 ," 00:42:20 yes 00:42:21 you can 00:42:24 and you can also do 00:42:42 create foo ," xxxx" ;uses count 00:42:46 ok, what would foo leave on the stack? adr of counted string? 00:42:47 erm no you cant 00:42:52 erm im fucking up here heh 00:42:57 address of a counted string 00:43:07 ok, that's the behavior of c" 00:43:27 so c" and s" are both defined in ans 00:43:40 i KNOW s" is because E.R. said so heh 00:43:52 one looks like it compiles a counted string 00:43:53 s" 00:43:57 the commitee member could *NOT* come to consensus about how string literals should be done ... but *DID* agree that strings literals were neccessary in any forth 00:44:04 c" looks to me like it compiles a counted CHAR ugh 00:44:15 they are 00:44:31 im also haveing a little difficulty with them - not with what needs to be defined 00:44:37 but with how i should name the words 00:44:48 ok, so they were forced to compromise by providing both methods with word names that weren't commonly used ... s" and c" 00:44:53 i think some of the string words are very badly misnamed 00:45:08 erm s" isnt in isforth 00:45:12 tho it mignt be later 00:45:25 ive never heared of c" but i already dislike it 00:45:35 it sounds likke its compiling a counted character 00:45:36 ugh 00:45:38 they also *EXPLICITLY* said, in an appendix, that they expected the app writer to name their favorite behavior into " and then write their apps as they always had 00:46:04 ,' compiles an uncounted string 00:46:08 it allows things like 00:46:09 yes 00:46:10 create foo 00:46:20 'f' c, 'o' c, 'o' cm 00:46:24 to be compiled as 00:46:25 create foo 00:46:28 ,' foo' 00:46:33 much neater 00:46:37 yep 00:46:45 i invented ,' but im sure it was a re-invention 00:47:00 its too obvious heh 00:47:21 create blah ," fu" 10 c, ," bar" 10 c, 00:47:53 that would compile 2 c, 'f' cm 'u' c, 10 cm 3 c, 'b' c, .... 00:47:54 win32forth uses ," with the same behavior (or nearly the same) 00:48:06 i sort of stole ," from fpc 00:48:33 anyway, that's why there's two string words in ANS, and why they're both oddly named 00:49:11 i am tempted to use the same names 00:49:13 " 00:49:14 ," 00:49:14 c" returns the address of a byte-counted string, and is deprecated for its limitations 00:49:15 s, 00:49:23 etc but redefine their behaviour 00:49:31 well 00:49:37 s" returns both the address and the length of the string literal 00:49:43 maybe they will stay the same - but " is 00:49:50 s" will NEVER be defined in isforth 00:49:51 EVER 00:49:59 ok, that's fine 00:50:03 nothig to stop users from defining it tho heh 00:50:06 jeez! have a cow on me 00:50:11 nonon heh 00:50:13 i wasnt heh 00:50:20 i was just being emphatic :) 00:50:42 if you have a counted string, then there's a headache about whether to store the count in 8, 16, or 32 bits 00:50:54 ANS didn't *EVEN* want to get into that 00:50:55 i think its one of the most fucked up words there is 00:51:01 after postpone 00:51:06 8 bits 00:51:08 final 00:51:15 8 bits is traditional 00:51:25 but not enough for many purposes 00:51:42 tz had a word that defined 16 bit counts too 00:52:12 yes - for straight ascii 256 byte limits are fine 00:52:24 now, s" is expected to store a counted string internally, but it's responsible for fetching the size (regardless of how many bits its stored in) ... problem solved 00:52:31 but who says a string is always going to be PURE ASCII 00:52:47 right 00:53:08 a coder might include entire web pages as literals 00:53:31 very fast, but likely to exceed 256 characters 00:54:05 i neway i realy need to go zzz 00:54:11 errr if he did he is silly heh 00:54:20 why silly? 00:54:45 he should store them as files heh 00:54:45 not embed them into his source files 00:55:02 It might get burned into the ROM of an internet appliance 00:55:06 the source might not chanve in 4 or 5 years 00:55:06 but the web info sure will 00:55:38 fine. burn sources into address 0xa000 00:55:46 burn html into 0xb000 00:55:50 no, you're right, not all pages are static, those will need some additional processing to embed the variable data 00:55:55 the two items are seperate 00:56:19 my cisco router, for example, has a web interface burned into flash 00:56:25 unless you code the html as forth words to output htm;l\ 00:56:38 yes, that is what I would probably do 00:56:59 my point is - html is not forth. forth is not html - th soruces for each should be cinsidered seperate entities 00:57:23 they are building web interfaces into everything these day7s heh 00:57:27 in fact, I'd probably use CREATE and ," and c, to build a 32bit counted string 00:57:35 instead of telnetting to the router you now BROWS it 00:57:35 much nicer 00:57:43 yep 00:58:16 which means the logical place for the rather large literal data that makes up the pages is in the dictionary, rather than in a file 00:58:25 ,' 00:58:25 ," compiles an 8 bit counted string 00:58:36 oh, ok 00:58:38 oohh 00:58:38 ."" ?? 00:58:38 ,"" xxxxxx " 00:58:51 a 23 bit ounted string ? 00:58:51 . 00:59:04 erm ... never seen that word, but you get the idea 00:59:04 ," xxx" 8 bit ,"" xxxxx" 16 bit ? 00:59:13 or ,2" xxxxx" 00:59:19 create a header, store a 0 CELL for a count 00:59:34 im trying to think of good names for these words 00:59:36 then just add as many chars as I want 00:59:45 thats why they arent defined yet heh 00:59:47 I'd lay data with ," 00:59:57 nono string compile words should nhot create word headers 00:59:58 most forths I've seen do 01:00:09 I didn't say they should 01:00:10 create foo compile string ehre 01:00:10 hmm 01:00:31 right ... create foo ," lay some uncounted literal data" 01:00:36 erm nooo i told you hehe 01:00:45 ," already compiles an EIGHT BIT count! heh 01:01:04 ,' <-- no count byte in this string' 01:01:06 yes, that's fine ... although it's not what I expected it to do 01:01:20 if that's the case, why not use c" ??? 01:01:26 double qoute - or single quote 01:01:35 oh, ok 01:01:37 ," compiles an 8 bit count in every forth ive looked at 01:01:48 i had to INVENT ,' 01:01:48 er 01:02:00 because c, is CHAR COMMA 01:02:00 so c" is going to compile a charactrer string ? 01:02:13 its a fucked in the head name :P 01:02:24 no win32forth has the same functionality, though I can't remember the name ... which means that FPC did too 01:02:29 see. this is my biggest critisism with todays forths 01:02:35 heh 01:02:38 whoever thinks these names up has their head up their ass heh 01:02:43 nope 01:02:53 fpc had no words to compile an uncounted string 01:02:58 you mean ANS? 01:03:02 other than 'x' c, 'y' c, 01:03:24 I've already explained that they deliberately chose names for the string literals that nobody else would use 01:03:25 there is one thing that ive not looked at in isforth 01:03:25 how the fuck does fpc do a ' ' literal heh 01:03:37 a 0x20 01:03:45 in isforth i get tick tick heh 01:03:53 in ans, it's SP I think 01:04:06 want me to check? 01:04:16 i.e. that nobody else would ever think of calling them and would therefore never remember. 01:04:28 ' badname alias goodname 01:04:35 nah 01:04:39 bl 01:04:49 that's exactly what ANS wanted everyone to do with the string literals 01:04:55 32 constant bl 01:05:07 as far as I know, that's the only place ANS had to choose goofy names 01:05:11 but if i allow 'a' 'b' 'c' i realy should figure out a way to allow ' ' 01:05:11 heh 01:05:39 postpone is a fucked up word and a fucked up name 01:05:41 yep, it's BL 01:05:49 nuff sed :P 01:05:51 * MrReach sighs. 01:06:12 i do have to agree tho - thats a valid use for the word alias 01:06:13 I'm not versed in it well enough to comment, really 01:06:24 fpc overused alias 01:06:28 the funny part is that ANS doesn't define ALIAS 01:06:36 ' [bp+si] alias [si+bp] 01:06:43 or anything like ALIAS 01:07:02 ' [bp+si] alias [rp+si] 01:07:12 ' [bp+si] alias [si+rp] 01:07:21 most forths have it, of course, but it's not expected to be uniform across platforms 01:07:36 he did that for every single possible combination of every register pair 01:07:44 yes, I remember 01:08:18 alias in forth is like goto in c 01:08:28 its OK... if you dont overuse it 01:08:34 heh 01:08:41 goto is evil in any language 01:08:51 its good for changing CONTEXT 01:08:51 like 01:09:02 'is' is an alias for !> 01:09:15 10 store-to some_var 01:09:16 --- quit: davidw (Read error: 104 (Connection reset by peer)) 01:09:23 --- join: davidw_ (~davidw@ppp-178-10.25-151.libero.it) joined #forth 01:09:35 actually, zimmer included those aliases at the request of one of the users of his system, via email 01:09:35 ' someword IS some-defered 01:09:35 its not 01:09:48 it just doesnt allow for very structured code if its overused 01:10:09 zimmer should have told that user to go to hell :P 01:10:16 or define them himself if he wanted that sort of bullshit 01:10:23 heh 01:10:48 is and store-to (!>) have identical execution 01:10:48 but different contexts 01:10:56 so the alias is good 01:11:02 he probably figured, "If one user has probs remembering the register order, many others probably do,too, and it's a waste of time to have to look them up" 01:11:16 aliasing every single combination of r1+r2 is BAD 01:11:33 yes, !> was aliased to IS after the ANS team sent him a threatening email 01:11:37 alwasy base first 01:11:48 bx+some-red 01:11:48 or 01:12:00 bp+some-red 01:12:12 threatned what ? 01:12:24 hell if I know 01:12:52 they were upset that he wouldn't make such a small change to be compliant, though 01:13:38 the cool part is that he now has LOCAL-ALLOCATE: 01:13:49 i always critisised tz for having "feature-itis" 01:14:00 maybe it wasnt HIS FAULT!!! 01:14:00 errr 01:14:02 which allocates a cell in the return stack as the word is entered, and you can use IS on it 01:14:10 wtf is that > 01:14:51 AND you can use the return stack freely while in the word, and it cleans up automatically on the way out 01:14:59 er 01:15:01 almost the way C does it internally 01:15:45 call fucntion. open local stack fram. do function. clean local stack frame. clean parameters off stack 01:15:53 yea. very efficient 01:15:53 lets add that to forth :P 01:16:00 i realy do gtg hehe 01:16:17 im falling assleep at the kbd 01:16:18 no, much more efficient, because params are on a different stack 01:16:26 the problem is you always have shit tahts interesting to chat about heh 01:16:26 dammit! 01:16:27 oh, ok 01:16:53 heh, I'm known for keeping people up 'til 4 in the morning talking about stupid shit that's interesting 01:16:58 yea - i know 01:17:02 even IRL 01:17:22 :) 01:17:30 --- nick: davidw_ -> davidw 01:17:31 you sleep well 01:17:34 irl would be fine 01:17:34 heh 01:17:49 naw .. 01:17:55 nite nite 01:18:10 then I'd want to drink, then I stop talking about forth and start talking about souls 01:18:15 --- quit: I440r (": sleep bed go tuck light off ; immediate") 01:20:30 --- nick: MrReach -> MrGone 01:39:00 --- join: Fare (fare@samaris.tunes.org) joined #forth 02:33:13 --- quit: Fare ("61453") 03:41:07 --- join: rob_ert (~robert@h237n2fls31o965.telia.com) joined #forth 03:56:31 --- join: Fare (fare@samaris.tunes.org) joined #forth 04:01:52 --- nick: Soap` -> SoapZZzz 05:53:40 --- join: onetom (tom@adsl52032.vnet.hu) joined #forth 05:54:55 --- quit: davidw (Remote closed the connection) 06:47:24 --- join: herkamire (~jason@wsip68-15-54-54.ri.ri.cox.net) joined #forth 06:51:29 ok, new I want a ?do :) 06:51:43 s/new/now 06:52:49 I finally realized that it's just 1 extra branch (and the branch doesn't have to execute every time through the loop) 07:40:03 --- quit: Fare ("Connection reset by pear") 09:00:32 oh, I just wrote the most useful piece of code ;) 09:00:44 = if true exit then false exit 09:01:13 I'm so proud ;) 10:06:23 --- join: tathi (~tathi@wsip68-15-54-54.ri.ri.cox.net) joined #forth 10:08:37 herkamire: what did you name that snippet of code? 10:08:44 ?exit ? 10:13:58 herkamire: actually, if you want to save some clocks, youcan do that using math functions only (flow control tends to be quite a bit slower) 10:15:09 in fact that whole phrase is redundant ... instead use <> ; 10:15:22 the exit is unneccessary 10:16:01 in both cases it exits, so just leave the (properly formed) flag on the stack at the end of the def 10:17:08 herk: how many C primitive does your forth have now? 10:18:33 can it do syscalls and load arbitrary shared libs at runtime? 10:18:38 yet? 10:29:10 yet? 10:29:20 no, and I don't think it ever will 10:29:54 ok, why? 10:30:19 70 words written in C 10:30:27 I thought it was about half that, but I just checked 10:30:46 cool, not a completely minimalist forth, then 10:31:13 I didn't make a way for C to call words defined in forth 10:31:24 the syscalls might be a bit tricky, I recall that gnu C uses a macro for them that lays some bytes inline 10:31:50 but loading an arbitrary library is incredibly easy 10:32:26