00:00:00 --- log: started forth/13.01.18 00:10:34 --- quit: ASau (Ping timeout: 272 seconds) 00:42:49 --- quit: proteusguy (Remote host closed the connection) 00:54:17 --- join: epicmonkey (~epicmonke@188.134.41.175) joined #forth 00:54:17 --- mode: ChanServ set +v epicmonkey 01:02:16 --- quit: bjorkintosh (Ping timeout: 264 seconds) 01:08:16 --- quit: epicmonkey (Ping timeout: 240 seconds) 01:14:25 --- join: bjorkintosh (~bjork@ip68-13-229-200.ok.ok.cox.net) joined #forth 01:14:25 --- mode: ChanServ set +v bjorkintosh 02:10:46 --- join: epicmonkey (~epicmonke@host-224-58.dataart.net) joined #forth 02:10:46 --- mode: ChanServ set +v epicmonkey 03:46:49 --- join: proteusguy (~proteusgu@ppp-58-8-100-194.revip2.asianet.co.th) joined #forth 03:46:49 --- mode: ChanServ set +v proteusguy 04:09:57 --- join: JDat (9f9411b5@gateway/web/freenode/ip.159.148.17.181) joined #forth 04:09:57 --- mode: ChanServ set +v JDat 04:43:57 it would seem that it might be more useful for words to have separate immediate and run-time definitions 04:52:05 why? 04:54:28 --- quit: proteusguy (Ping timeout: 264 seconds) 04:54:44 protist: I think Forth has been slowly moving away from state smart words. 04:55:58 You could have two separate vocabularies. One for compiling, one for interpreting? 04:59:30 impomatic: yeah...i am thinking about that 04:59:41 impomatic: it really would simplify things 05:00:21 impomatic: as it stands, making control flow words seems sloppy 05:00:40 --- join: I440r (~zhiming@167.sub-70-194-68.myvzw.com) joined #forth 05:00:40 --- mode: ChanServ set +o I440r 05:00:41 impomatic: maybe there are better ways...but i have defined special behavior in my assembly code 05:00:54 warning: gcc/cc1-checksum.o differs 05:01:11 anyone know why my emerge of gcc is failing on a checksum compare between stages 2 and 3? 05:01:17 argh wrong channel 05:01:33 impomatic: begin would be fairly simple with a normal immediate....but until has both immediate and run-time behavior 05:02:20 until is an immediate word that compiles the conditional branch back to the mark left by begin 05:02:42 --- join: tathi (~josh@dsl-216-227-94-171.fairpoint.net) joined #forth 05:02:42 --- mode: ChanServ set +v tathi 05:02:48 hi tathi! ltns! 05:02:55 Hi I440r 05:02:57 How goes it? 05:03:01 I440r: i suppose then i have a conditional branch word in asm...i suppose that makes more sense 05:03:29 well that depends lol - if i get the contract i just interviewed for all is fine :) 05:03:34 * tathi goes to look at the logs to see what we're talking about... 05:04:03 been working ALOT on isforth, fixed a bunch of bugs and am implementing the pulldown menus... just gotta fix a few bugs in that :) 05:04:07 Cool 05:04:16 BEGIN saves an address. UNTIL retrieves the address and compilers an OBRANCH back to it. BEGIN and UNTIL only make sense within a colon definition. 05:04:40 impomatic: i have begin and until both implemented at the assembly level 05:04:46 protist, : begin compile dobegin impomatic: but my interpretter has no concept of immediate, yet 05:05:09 the only reason for "dobegin" is so the decompiler knows that there is a begin loop here 05:05:32 : until compile ?until ?until is an alias for ?branch 05:05:58 protist, why not? that complicates the issues big time 05:06:13 how does ; work if thers no immediate? 05:06:24 : ; compile exit [ ; 05:06:28 i didn't want to read how it is normally done until i implemented it myself 05:06:43 i have if,else,then,begin,until all working....using assembly 05:06:57 protist: so how do you stop compiling and go back to interpreting? 05:07:07 well immediate words are flagged as being immediate by having bit 7 of their NFA's count bit set 05:07:13 --- join: proteusguy (~proteusgu@ppp-58-8-230-61.revip2.asianet.co.th) joined #forth 05:07:13 --- mode: ChanServ set +v proteusguy 05:07:29 tathi: colon is a word that has an assembly definition haha 05:07:31 : ; compile exit [ ; immediate (forgot to make it immediate lol) 05:07:36 tathi: in mine, anyways 05:07:47 protist: but how does it know when the definition ends? 05:08:31 tathi: ill show you, sec 05:08:36 k 05:08:39 protist, there are forths that have no concept of immediate but i think thats about as useful as the GNU itterative quicksort :) 05:09:03 having immediate words simplifies things greatly 05:09:18 tathi hows the farm work going? 05:09:20 still doing that? 05:09:26 I440r: yup. 05:09:31 Strange weather. :) 05:09:41 global warming! 05:09:42 lol 05:09:51 tathi: http://ideone.com/GLe6GD 05:10:00 A couple days ago it was in the 40s (Fahrenheit), which is more like March weather, and now we're back to -6 this morning... 05:11:01 I440r: we got way behind in the fall; didn't pull the tomatoes until after Thanksgiving, so we're short on salad production right now. But otherwise things are good. 05:11:21 protist: oh, colon parses the input. 05:11:33 protist, i think i need to explain quit and interpret 05:11:40 tathi: for the duration of the definition, yes :) 05:11:49 I440r: hmmm 05:11:49 some forths use ] to do the parsing too 05:12:20 quit is forths inner interpreter. quit is basiclaly begin interpret again - i.e. an infinite loop 05:12:28 protist: Yeah, you can do that. The trouble is then it's difficult to write words which use colon to compile other definitions. 05:12:39 if there is an error during interpretation the word abort jumps back to quit after cleaning up the stacks 05:12:48 unless you provide the factor which just starts a definition and doesn't parse input. 05:13:28 consider : noop ( "name" -- ) : postpone ; ; 05:14:29 interpret is a loop that parses the next space delimited token out of the input stream, searches the dictionary for it and then either compiles or executes it depending on "state" 05:14:56 tathi why not just : noop ; 05:15:21 I440r: because it's an example to demonstrate what's wrong with a parsing : or ] . 05:15:28 it defines a noop word 05:15:32 in fact i think your definition for noop would be wrong but oh :) 05:15:47 your definition for noop would also do ] making it wrong :) 05:15:50 but i get ur point 05:16:27 if your compilation state is a parsing word, then trying to do `noop foo` won't define foo as `: foo ;`, it will eat input until the next semicolon in the input stream. 05:17:06 oh your noop definition should be called noop: or something. make the next token a NOOP word 05:17:16 I440r: oh, I've been sporadically working on something you'll probably hate: a Forth dependency/package manager. ;) 05:17:18 tathi: i can do : noop ; 05:17:35 : noop: create compile ; ; 05:17:40 noop: foo 05:17:48 protist: right, but what if you want to do a word that compiles other definitions? 05:17:57 --- join: foxes (~fox@124.193.192.54) joined #forth 05:17:58 --- mode: ChanServ set +v foxes 05:18:02 tathi a forth dependency package manager for FORTH stuff? 05:18:08 or for some distro ur making :) 05:18:10 protist: like a structure field word, which takes an offset on the stack and compiles a word which adds that offset to an address. 05:18:17 I440r: no, for Forth stuff. 05:18:32 tathi: i am not sure what you are wanting haha 05:18:41 so if you include foo.h and foo.h needs bar.h and futs.h they will automatically be included? 05:18:48 agreed. i dont like that :( 05:18:50 :) 05:18:51 even 05:18:52 tathi: i have defined create...if that would illustrate how mine works? 05:18:57 it encourages you to not know WTF is going on 05:18:57 I440r: yeah. :) 05:19:17 tathi: : create get_lexeme here dup ep, write dp ! here ep, con_type , here cell + , ; 05:19:42 tathi i totally reworked my memory manager, its way more organized now, not all in 1 file. much easier to read the sources 05:19:57 also did major surgery on the curses code including fixing a bunch of stupid bugs lol 05:20:15 protist: sure. Does this make sense? : +field ( offset u "name "-- offset+u ) create over , + DOES> ( addr -- addr' ) @ + ; 05:20:17 need to rewrite see and the debugger. they are currently broken 05:20:50 protist: used as 0 1 cells +field foo 1 cells +field bar constant sizeof-struct 05:20:57 I440r: cool, I'll have to take a look 05:21:26 --- quit: Fox78 (Ping timeout: 244 seconds) 05:22:15 its not released yet 05:22:35 need to finish up the pulldown menus which are looking real good except when it glitches :) 05:22:41 ok 05:23:12 tathi: i am not sure what you are wanting that to do 05:23:28 tathi: i don't see how it relates to `:' 05:23:59 protist: in a more standard forth, you can write something similar in which the definition of +field uses : and ; instead of CREATE 05:24:18 protist: in yours, you can't. 05:24:47 tathi: ah mine could probably do that pretty easilly 05:25:10 protist: yeah, it's not a huge deal, you just need a separate word which only starts a new definition and doesn't have the loop part. 05:25:17 tathi like my m: definition. m: has : inside it 05:25:46 protist, can you do [ x y z ] inside your compile loop? 05:26:03 I440r: yes 05:26:21 I440r: i can do : five [ 5 ] literal ; 05:26:52 I440r: or : four [ 2 2 + ] literal ; 05:27:58 protist: in a normal Forth, there is one main interpret loop, and : and ; (or actually [ and ]) just set a flag which changes the behavior of that loop from compile to interpret. 05:28:20 protist: then you can use : and ; do do things like this rather than having separate words. 05:29:09 hmmm 05:29:11 protist, the main loop is called quit. which calls interpret which repeatedly calles defined (see if next space delimted token is defined) until it runs out of input (end of line et) 05:29:36 if defined finds the word it then either compiles it or executes it depending on the current state 05:29:57 so forth just keeps parsing each token out of the input stream over and over till theres none left 05:30:05 protist: your choice: it's a fairly minor difference. The only real problem is that it may confuse people coming from other Forth systems. 05:30:07 and as its doing this it is being switched into and out of compile state 05:30:11 : blah ... ; 05:30:24 protist: but if you're not planning to have other users, it's totally up to you. :) 05:30:37 forth parses colon. colon parses the next token and creates a word with that name. colon then switches to compile mode. 05:30:39 if i wrote an outer interpretter...it would work like : outer begin get_lexeme dup lookup if lookup execute else number then 0 until ; 05:30:49 i tried something like that, and it works 05:31:01 interpret then continues to parse each token and because were now in compile mode it compiles each token instead of executing it 05:31:23 ; however is an immedite word so interpret EXECUTES it. ; compiles an exit then switches back to interpret mode 05:31:42 thers one loop and it does all the input parsing and then either executes or compiles 05:32:06 this looks more complicated but its much much simpler really... it also allows alot of "clever stuff" tahts not possible otherwise 05:32:29 I440r: interesting...i will likely make more Forths once i get a firmer grasp on this one 05:32:42 oh.. if interpret parses something thats NOT in the dictionary 05:33:02 it checks to see if its a number. if its a number it either puts it on the stack or compiles it as a literal 05:33:19 OK, I'd better get to work. 05:33:25 later tathi :) 05:33:29 tathi: cya 05:33:31 later 05:33:36 --- quit: tathi (Quit: leaving) 05:34:09 protist, tathi is the guy that ported isforth to PPC but he is farmer joe now not geek master so the ppc version isnt being maintained :) 05:34:15 I440r: i may add in ?branch and such so i can make outer definition for control flow words 05:34:40 you need branch and ?branch and ?branch seems backwards... it branches on false not on true 05:34:50 so IF is a ?branch round the true part 05:35:10 : if compile ?branch >mark ; (where TO MARK marks a forward branch) 05:35:16 yeah...i have all those control words defined at the assembly level 05:35:55 i am thinking I'll add in the concept of immediate, then generalize with 0branch and ?branch to make outer definition prettyness and compactness 05:36:09 i would highly recommend you become familiar with how isforth does it... 05:36:51 i am trying my hand at it my way before i look to the normal ways, so i can see where they differ...I will take more from this that way 05:37:01 this won't be the only Forth i write lol 05:37:03 isforth has 0branch but it doesnt use it at all lol 05:37:10 i shud remove that primitive, 05:38:01 : until literal compile ?branch ; 05:38:06 something like that 05:38:08 i know but there are some concepts that your not familiar with yet. i USED forth for about 6 or 7 years before i even started to look at how part of it worked. i dint write isforth till i had been using forth for over 10 years :) 05:38:19 as an immediate 05:38:23 not literal 05:38:32 : until compile ?branch and yes its an immediate 05:39:03 i use the normal stack to pass control destinations 05:39:10 so i could use literal 05:39:19 : begin here ; 05:39:37 or hmm 05:39:39 thas a very simple definition for begin. all it does is put the address to be looped back to onto the parameter stack 05:39:48 : again compile branch , ; 05:39:53 yeah 05:40:04 so again compiles branch and the address to branch back to 05:40:07 --- join: fantazo (~fantazo@213.129.230.10) joined #forth 05:40:07 --- mode: ChanServ set +v fantazo 05:40:39 i have all of these on the assembly level, even `else' lol...so i understand the mechanisms...these words are cleaner, though 05:40:48 : if compile ?branch here 0 , ; 05:40:56 the 0 , is a dummy branch target 05:41:08 : then here swap ! ; 05:41:31 all of those are immediate of course 05:41:36 why the 0? 05:41:47 bcause branch has a target... but we dont know what it is yet 05:41:53 so we compile a dummy branch target 05:42:02 ah yeah 05:42:06 ?branch is TWO tokens. 1 token for ?branch and one for its target 05:42:21 gotcha 05:42:45 so THEN does "here swap !" 05:43:12 i.e. back patch the ?branch 05:43:39 you run linux right protist ? 05:44:04 yep 05:44:25 my idea of how i was going to do ?branch was a little different 05:47:05 I440r: i have done all the same things, just not generalized, and in asm haha http://ideone.com/lBxgkd 05:47:56 my `else' comments may not be that great 05:49:17 :) 05:49:28 :D 05:49:47 i wanted to try my hand at it to really understand it before looking to standard implementations 05:50:32 gaining the speed of pure asm at compile time is hardly worth the cost in space...compile time need not be fast, anyways 05:51:02 so i will switch these things over, but this experience has been valuabe :) 05:51:18 valuable* 05:54:28 protist, i disagree, compile time is wasted time 05:54:32 my fears of possibly being token-threaded were unfounded, while i do have a separate table, it is there so that i don't have to dereference links in a linked list to traverse the dictionary 05:54:39 emerge open-office or mozilla-firefox for an example 05:55:16 what is emerge? 05:55:27 its gentoo "install".. 05:55:35 gentoo downloads, extracts, compiles and then installs 05:55:49 isforth just compiled 248498 in .02 seconds 05:56:20 i have an entry list where the entries are of known size so i can use address arithmetic to move through it rather than a linked list 05:56:29 thats 242k plus 700 ish bytes 05:57:02 you cant do vocabularies or hashing of names that way 05:57:15 you need to do linked lists. 05:57:31 hashing is too fancy for my first real foray into assembly haha 05:57:37 :) 05:57:42 next time i will experiment :D 05:57:52 i have learned so much asm doing this 05:58:03 hashing of forth names is simple. take the length *2 + the first char of the name *2 plus the 3rd char of the name if it exists 05:58:07 i read most of a book on asm, but hadn't applied it before this 05:58:14 thats basicaly isforths entire hashing algorithm 05:58:32 i have done hashing in C, before :D 05:58:36 i know the mechanism 05:58:55 most asm books these days are utter trash though. 3/4 of the book is spent teaching you the binary, octal and hex and the differences between masm, tasm and spazzum 05:59:12 then yu get "." <--- about that much of the book dedicated to actual assembler coding 05:59:29 i used Professional Assembly Language, it was /fairly/ good :) 06:00:00 it was more aimed at C programmers though, so it didn't do data structures 06:00:14 get "The 8086 book, includes the 8088). its OLD 16 bit dos asm but its a billion times better than any book published on asm since the 1980's 06:00:47 cool 06:00:51 i may look into that :) 06:01:09 its out of print lol... you can probably get it on amazon tho 06:01:42 what about Michael Abrash's Graphics Programming Black Book? 06:02:31 to learn gfx stuff or to learn asm 06:03:09 asm 06:03:22 i heard in spite of the title, the first third is golden for asm 06:03:40 well the guy is a freeeeeeeking genius lol 06:03:53 if you truely want to learn asm again... read my sources :P 06:04:00 hearing that from you means something 06:04:09 which guy is a genius?...Abrash? 06:04:15 yes abrash 06:05:32 it is lame how meeting with a girl is always a 50/50 chance scenario lol 06:05:56 they are like "yeah, sure!", but it is never a sure thing 06:18:38 is the word immediate always used right after a definition? 06:18:53 I440r: ^ if so it would be easy for me to define 06:22:46 --- quit: rabenauge (Quit: Ex-Chat) 06:24:24 --- join: rabenauge (~sag@88.130.166.126) joined #forth 06:24:25 --- mode: ChanServ set +v rabenauge 06:27:06 yes 06:27:11 : foo ... ; immediate 06:27:39 the word create stores the name field address of the most recent word name in a variable called last 06:28:40 last c@ $80 or last c! (set bit 7 of the count byte of the most recently created word) 06:28:50 usually catually done as 80 last cset 06:28:58 where cset is a read/modify/write 06:29:46 and FIND - the word that searches the dictionary will return 0 = not found. -1 = found but the word is not immediate or... 1. word found and is immediate 06:30:07 -1 and 1 are both non zero so are "True, word found" as opposed to "false word not found" 06:30:32 after that you can do "state @ xor if , else execute then 06:30:46 erm or do i have the if/else parts backwards lol 06:31:05 state = 0 = interpret. state = -1 = compile 06:32:09 so if find returns -1 (word found) and state is 0 (i have the if/else parts backwards for sure) we need to execute not compile 06:32:16 --- quit: proteusguy (Ping timeout: 264 seconds) 06:33:13 err i might be getting this all wrong. you xor the result of find with "State" to determin if you compile or execute based on immediate/state 06:33:21 look at how isforth does it :P 06:33:33 hehe 06:33:40 man i have a gay error 06:34:02 one of those errors that pretends it is on 600 lines when you compile 06:34:40 just find the first error and fix it. that will tell you where your next real error is 06:35:18 I440r: yeah 06:35:47 I440r: it claimes the first error is line 265...i guess i haven't changed much before there 06:36:00 OH 06:36:04 i didn't use -m32 06:36:05 haha 06:36:07 that is it 06:37:46 --- join: Nisstyre-laptop (~yours@oftn/member/Nisstyre) joined #forth 06:37:46 --- mode: ChanServ set +v Nisstyre-laptop 06:41:20 I440r: i now can make immediates :D 06:42:03 I440r: i made my own version of 0branch and such, that are slightly different 06:42:04 which is going to make your all in one compile loop more complicated 06:42:20 but if it works its a good start 06:42:26 well 06:42:43 its not that the way its usually done is LESS complicated... its just more organized 06:43:13 : immediate imm_type ep @ cell - @ ! ; 06:43:17 :D 06:48:44 : if con_type , here cell allot compile ?branch ; 06:49:37 immediate 06:50:29 works :D 06:51:03 much, much, more elegant and compact 06:51:25 s/much, mo/much mo/ 06:51:44 oh shit...it might be using the old if 06:51:53 i'll change the asm and test :D 06:52:51 seg fault hmm 06:53:04 ah 06:58:22 lol love those 06:59:10 --- quit: cratuki (*.net *.split) 06:59:15 --- join: cratuki (~Craig@leaf.vm.bytemark.co.uk) joined #forth 06:59:15 --- mode: ChanServ set +v cratuki 07:17:02 I440r: does this look right to you? : compile ' , ; immediate 07:17:08 brb drink 07:17:52 not quite 07:18:36 thats [compile] 07:18:49 [compile] takes the next token out of the input stream and compiles it 07:19:00 compile takes the next token out of the EXECUTION stream and compiles it 07:19:34 so if you have compile foo 1 2 3 do-something at run time COMPILE runs,,, fetches "foo", commas it the we continue executing from 1 2 3 07:20:04 so : compile r> dup cell+ >r @ , ; 07:20:19 that sort of thing 07:20:42 and compile is not immediate but [compile] is 07:21:05 hmmm 07:21:17 if you have : blah .... compile foo .... ; 07:21:29 thers an XT for "compile" followed by an xt for "foo" 07:21:36 when forth executes the word compile 07:22:19 it pulls its return address off the return stack (getting the address of the xt for foo).. 07:22:55 the dup is needed because ewere going to use this address twice sort of. we first add 1 cell and put it back on the return stack so when we return we return to the xt after the xt for foo 07:23:06 in your definition of if, i thought you used compile ?branch 07:23:21 we do 07:23:30 : if compile ?brahch here 0 , ; 07:23:33 this compiles down to 07:23:35 if: 07:23:40 dd compile 07:23:44 dd ?branch 07:23:46 dd here 07:23:48 --- join: MayDaniel (~MayDaniel@unaffiliated/maydaniel) joined #forth 07:23:48 --- mode: ChanServ set +v MayDaniel 07:23:57 : compile ' , ; immediate seems right for that? 07:23:58 dd (lit), 0 07:24:03 no 07:24:33 oh...that would be compiling at the wrong time 07:24:33 ok.. a colon definition is just an array of "execution tokens" or xt's 07:24:39 an xt is just the address of a word 07:24:41 right 07:24:45 i see now 07:24:49 lost in all the immediates hehe 07:24:56 it can get confusing at first 07:25:06 it seems squirly but... its not once you get your head straight :) 07:25:16 gotcha 07:25:33 i am getting things close enough to normal that i can talk to you, which is good haha 07:25:37 : compile r> dup cell+ >r @ , ; 07:25:46 : [compile] ' , ; immediate 07:25:50 TICK is a parsing word 07:25:59 so bracket compile is a parsing word. 07:26:06 --- quit: fantazo (Quit: leaving) 07:26:27 i.e. it compiles the next token out of the input stream... which is itself usually immediate 07:26:40 very nice :D 07:26:46 in fact the only reason for [compile] is to COMPILE an immediate word. if you didnt [compile] it it would execute 07:26:57 : fud ." fud" cr ; immediate 07:27:00 : blah fud ; 07:27:12 fud would not be compiled into the definiton for blah because it would execute at compile time 07:27:19 : blah [compile] fud ; 07:27:31 --- quit: Nisstyre-laptop (Remote host closed the connection) 07:27:35 [compile] un-immediatifies a word 07:27:45 makes sense 07:28:05 : blah compile [compile] fud ; 07:28:10 what does that do 07:28:14 at compile time AND at run time 07:28:29 i need to better understand compile first 07:28:35 i get [compile] 07:28:43 that compile definition is odd 07:28:52 first tell me what it does exactly 07:28:55 ok compile is not immediate so the XT for compile is just stuffed into the definition for blah 07:28:57 ok 07:29:25 if you have a forht word that has XXX YYY and ZZZ inside it. forth points to each one and fetches that xt and executes it 07:29:32 you understand how (lit) works? 07:29:43 i don't have (lit) 07:29:53 ok. LITERAL is an immediate word 07:29:59 i have literal, though 07:30:06 : literal ( n1 --- ) compile (lit) , ; 07:30:06 but defined in asm 07:30:20 literal is a compiling/immediate word 07:30:30 (lit) is the primitive for literal, the RUN time action 07:30:55 so a number in forth is compiled as 07:30:57 dd (lit) 07:31:00 dd 123 07:31:20 now immagine forths IP is pointing at the xt for (lit) 07:31:41 forth fetches the xt its pointing to. it then advances IP to the nxt xt and then jumps to the one it fetches 07:31:46 its like an instruction fetch in asm 07:31:46 : literal uv @ , con_type uv, uv, ; immediate ....would be something like that for me 07:31:51 fetch, advance, execute 07:32:06 whe (lit) executes IP now points to the 123 07:32:23 it fetches the 123 pointed to by IP and advances ip to the next xt 07:32:38 what is uv 07:32:54 user_variable memory pointer 07:33:07 oh i dont have user variables 07:33:10 what is con_type 07:33:10 i put constants in there too 07:33:42 for me, constants are like 3----value 07:33:58 not sure what u mean by that 07:34:03 i have to have a reference to that, then execute will push the value on that stack 07:34:25 ok 07:34:27 stop 07:34:27 3 is the constant type 07:34:31 stop stop :) 07:34:34 :) 07:34:49 ok there are TWO kinds of words in forth. high level code or asm code 07:35:17 each word has an NFA = name field address and a CFA, code field address 07:35:25 high level definitions also have a body address 07:35:31 i have 4 kinds right now: primitive, user_defined, constant, and immediate 07:35:41 for ALL words, the first thing in the word definition is CODE 07:35:46 lets look at the code for + 07:35:49 code + 07:35:51 pop ax 07:35:53 pop bx 07:35:56 add ax, bx 07:35:58 push ax 07:36:03 next <-- more on this later 07:36:10 all coded definitions end in NEXT 07:36:20 a COLON definiton looks like this 07:36:24 foo: 07:36:27 call nest 07:36:28 mine is indirect threaded, not direct :) 07:36:29 dd xx 07:36:32 dd yy 07:36:33 dd xx 07:36:36 dd exit 07:36:48 you can do the translation later 07:37:04 indirect is over complicated. i see ZERO use for indirect threading 07:37:16 compact code 07:37:32 does not give more compact code 07:37:50 it also means next has to do TWO fetches to see what its executing 07:38:07 in the above code lets say we have 07:38:13 : bar foo ... ; 07:38:14 so 07:38:16 bar: 07:38:19 call nest 07:38:22 dd foo 07:38:23 ... 07:38:25 exit 07:38:34 lets say were executing bar and IP poits at the DD foo 07:38:53 forth does a read of where ip points to. advances ip. then jumps to foo 07:39:06 foo is ASM code. or at least the first part is 07:39:10 it does a call nest 07:39:34 nest pushes IP onto the return stack and then POPS the return address of the "call nest" into ip 07:39:45 so IP now points at the first token of foo 07:39:53 ok so how next works 07:40:16 in isforth IP is the SI register. you can fetch the data pointed to by SI with a "lodsd" opcode 07:40:23 taht fetches and advances SI 07:40:27