00:00:00 --- log: started forth/04.08.28 00:04:19 --- quit: arke (Read error: 110 (Connection timed out)) 00:08:56 --- join: arke (~chris@wbar8.lax1-4-11-100-111.dsl-verizon.net) joined #forth 00:10:21 --- quit: arke ("brb") 00:10:41 --- join: arke (~chris@wbar8.lax1-4-11-100-111.dsl-verizon.net) joined #forth 00:25:03 --- join: mur (~mur@kyberias.uiah.fi) joined #forth 00:26:57 hi 00:27:02 hi 00:28:46 what is eh up? 00:29:57 debugging :( 00:30:54 thats what you get for using C you loser! :P 00:31:02 no this is java 00:31:06 easier to debug than C 00:31:26 oh 00:31:27 :) 00:31:30 even easier? 00:31:32 Forth! 00:31:32 :) 00:31:50 even easier? factor :) 00:32:05 the problem here is not the language but the fact i wrote the code a long time ago 00:32:21 :) 00:33:16 http://cvs.sourceforge.net/viewcvs.py/jedit/jEdit/org/gjt/sp/jedit/textarea/DisplayManager.java?rev=1.92&view=auto 00:33:35 i dont trust links 00:33:38 that one file is more complex than an entire forth implementation 00:33:40 they might lead to goatse 00:33:51 and there's 100 of them 00:33:55 no, its a cvs link :) 00:34:17 JEESUS CHRIST THIS IS _HUGE_ 00:34:46 its about 1/50th of the whole program 00:35:17 i guarantee that this could be rewritten in the same language, but smaller, compacter, and faster 00:35:50 probably true, except for the faster part ;) 00:35:58 ack 00:35:59 :) 00:36:10 i optimized it intensly :) 00:37:02 but 00:37:07 if all this does is display text 00:37:10 why this much? 00:37:10 : 00:37:36 word wrap, folding, syntax highlighting... 00:37:37 --- quit: mur_ (Read error: 110 (Connection timed out)) 00:38:11 bad design... :) 00:38:20 how much memory do you sacrifice for speed? 00:38:29 :) 00:38:44 the overhead is 12 bytes per line of text 00:38:59 aha 00:39:06 there you go\ 00:40:38 try this 00:40:48 arke, much nicer: http://cvs.sourceforge.net/viewcvs.py/factor/Factor/Factor/contrib/mandel.factor?rev=1.4&view=auto 00:41:06 keep the lines in the display manager cache in a different format 00:41:20 as in, linked list, with a class describing attributes 00:41:21 the display manager has no overhead 00:41:28 such as color, behaviour, and the actual text 00:41:51 the per-line overhead is: 00:41:57 4 bytes: line start offset 00:42:09 2 bytes: fold level 00:42:18 2 bytes: number of lines on screen (from word wrap) 00:42:31 2 bytes: various flags & padding 00:43:29 i'd say employ a line cache system 00:44:22 to avoid per-line storage/ 00:44:22 ? 00:44:34 no 00:44:47 you still keep lines in a class 00:44:55 but it would be a linked list of items with attributes 00:45:06 what attributes? 00:45:22 color, type, behaviour, is it a word or whitespace?, etc. 00:45:43 none of that needs to be stored 00:45:47 just kinda keep it together like that. 00:46:27 including info on current text width for the wrap 00:47:02 if you change it (like, say you maximize it), just whatever. if you encounter an obsolete class, delete it. 00:47:19 should be trivial to convert from the linked list to ASCII 00:47:22 or display 00:47:29 and itll take up a bit more mem 00:47:32 i won't be making many changes to jEdit apart from maintanance. 00:47:34 but im sure itll be faster. 00:47:37 aah. 00:47:40 but hows my idea? 00:47:40 :) 00:47:52 its "good enough" right now, and i'd rather focus on more innovative stuffs. :) 00:48:35 :) 00:48:39 but i like that idea 00:49:00 dont go hunting for classes to delete them. leave them where they are, and delete them when you happen to encounter them. 00:49:02 then make an editor :) 00:49:14 i will 00:49:14 you mean instances of classes? (= objects)? 00:49:15 but 00:49:27 IT WONT BE A JAVA ONE! :) 00:49:33 yes, that would describe a line 00:50:05 including word wrap information 00:50:05 i'd like to do a factor GUI eventually 00:50:06 :) 00:50:43 i wish i didn't add word wrap 00:50:48 i never wanted it but users asked for it 00:50:56 and all it did was complicate the design *tremendously* 00:51:14 folding was a good idea, and didn't complicate too much 00:54:10 altho if you're making a forth editor, folding might not be worth it due to the horizontal nature of forth source 01:12:13 --- join: solar_angel (~jenni@MTL-ppp-157567.qc.sympatico.ca) joined #forth 01:18:08 well i fixed my bug. :) 01:18:10 good night. 01:18:46 night slava. 01:20:53 --- quit: solar_angel ("later.") 01:21:12 --- quit: slava ("Leaving") 01:41:54 --- join: proteusguy (~proteusgu@66.168.185.144) joined #forth 01:44:46 --- join: mur_ (~mur@smtp.uiah.fi) joined #forth 01:47:38 --- quit: mur (Read error: 60 (Operation timed out)) 02:25:59 --- quit: qFox ("meh ircii's netsplit detection appearantly sux ^^") 02:38:24 --- quit: mur_ (Remote closed the connection) 02:38:56 --- join: mur (~mur@mgw2.uiah.fi) joined #forth 03:08:12 * TomasuDlrrp is back (gone 07:04:52) 03:29:58 --- join: qFox (C00K13S@82-169-140-229-mx.xdsl.tiscali.nl) joined #forth 03:32:07 --- join: mur_ (~mur@mgw2.uiah.fi) joined #forth 03:42:48 --- quit: mur (Read error: 110 (Connection timed out)) 04:00:24 --- join: crc (crc@50-pool1.ras11.nynyc-t.alerondial.net) joined #forth 04:16:29 * TomasuDlrrp is away: dlrrp again 04:27:26 --- join: mur (~mur@mgw2.uiah.fi) joined #forth 04:39:59 --- quit: mur_ (Read error: 110 (Connection timed out)) 05:10:30 --- nick: TomasuDlrrp -> Tomasu 05:10:32 * Tomasu is back (gone 00:54:03) 05:46:36 --- join: FlamingRain (~Ecoder@c-24-129-95-254.se.client2.attbi.com) joined #forth 05:47:04 --- join: mur_ (~mur@mgw2.uiah.fi) joined #forth 05:59:17 --- quit: mur (Read error: 110 (Connection timed out)) 06:08:21 RetroForth/Linux now has working file I/O! 06:08:27 * crc is quite happy 06:19:16 --- quit: FlamingRain ("Leaving") 06:36:25 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth 06:38:58 --- join: FlamingRain (Ecoder@c-24-129-95-254.se.client2.attbi.com) joined #forth 07:06:00 --- quit: FlamingRain ("Leaving") 07:24:14 --- quit: tathi ("leaving") 07:31:24 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth 07:31:24 --- mode: ChanServ set +o Herkamire 07:32:06 Hi Herkamire 07:41:29 --- quit: tgunr (Read error: 104 (Connection reset by peer)) 07:42:09 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth 08:09:38 --- quit: tgunr (Read error: 54 (Connection reset by peer)) 08:10:21 --- join: tgunr (~davec@vsat-148-63-4-107.c001.g4.mrt.starband.net) joined #forth 08:10:44 --- quit: crc ("Time for bed... Goodnight All!") 08:25:10 --- join: mur (~mur@smtp.uiah.fi) joined #forth 08:37:16 --- quit: mur_ (Read error: 110 (Connection timed out)) 08:46:18 * Tomasu is away: dlrrp 08:46:23 --- nick: Tomasu -> TomasuDlrrp 09:22:00 --- join: mur_ (~mur@kyberias.uiah.fi) joined #forth 09:32:30 --- quit: mur (Read error: 110 (Connection timed out)) 09:34:30 hi crc :) 09:39:14 --- join: poing_ (~chuck@64-121-15-14.c3-0.sfrn-ubr8.sfrn.ca.cable.rcn.com) joined #forth 09:39:15 hmmm. sorta missed that one 09:48:15 argh. what happened to retroforth.org? 09:48:26 7.3 is due out today, no? 09:49:19 whup there it goes 09:54:39 --- join: FlamingRain (~Ecoder@c-24-129-95-254.se.client2.attbi.com) joined #forth 09:57:40 I440r_: hey d00d, you there? 10:03:58 He's probably d0ed. 10:04:13 Hmm.. 10:05:00 Funny. d0ed (död, Swedish for dead) and "dead" itself can both be written as hex numbers. 10:05:11 hehehe, cool 10:05:15 And a dead Swede < a dead Englishman (0xd0ed < 0xdead) 10:05:21 mov [i440r], word d0ed 10:05:31 Robert: :) 10:06:34 0xDEADBEEF 10:07:11 b00b babe, dead beef cafe, d0 da deed! 10:07:29 (poetry) 10:08:09 sortof a bad sign for your poetry if you actually have to tell people it's poetry ;) 10:12:14 it's ART dammit! 10:12:21 * poing_ straightens his beret 10:12:41 --- nick: poing_ -> poing 10:13:00 --- quit: FlamingRain ("Leaving") 10:13:11 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth 10:15:46 --- join: crc (crc@102-pool1.ras11.nynyc-t.alerondial.net) joined #forth 10:20:16 Still quiet in here... :-) 10:22:05 and you're ruining it! 10:22:30 hi 10:22:33 :) 10:22:33 Hi arke 10:22:44 whatcha doing? 10:22:48 is retroforth STC or DTC? 10:22:54 STC 10:23:00 yay. 10:23:02 I dislike DTC code 10:23:05 STC? 10:23:07 DTC? 10:23:13 subroutine/direct threaded code 10:23:15 ah 10:23:31 Working on RetroForth 7.3 right now 10:23:52 i created STC on roberts DTC forth :p 10:24:00 well, not true dtc i think 10:24:05 :-) 10:24:18 is all in assembly macros, so it should be pretty cheap subroutine threading 10:24:39 * crc is getting tired... 10:25:06 ran retro native under qemu, works flawlessly 10:25:13 It does 10:25:27 I can't get 'type' to work properly under qemu for Windows 10:25:31 maybe i can teach myself something about vga and get some graphics :) 10:25:41 :) 10:25:42 hey 10:25:47 works under qemu for linux tho? 10:25:52 there's no VGA driver in the current native release 10:25:57 I don't have qemu for Linux 10:25:58 arke, no im not here 10:25:59 :) 10:26:06 i remember offering l33t gfx for retroforth to crc if he implemented vm86 :) 10:26:08 I440r_: :) 10:26:08 tathi ya fix that math ?? :) 10:26:17 yah was thinking of learning vga stuff myself, maybe putting it in 10:26:27 got any pointers on where to start? 10:26:41 poing: yeah --- ralf browns interrupt list, int 0x10 10:27:10 ralf browns interrupt list is FSCKING HORRIBLE 10:27:12 lets 10:27:13 make 10:27:13 * poing is actually pretty new to such low level programming 10:27:16 ablahb'a 10:27:19 everything 10:27:20 Older RetroForth releases (think prior to 6.07 and lower) had decent 320x200 VGA driver 10:27:21 it's why i'm starting with forth :) 10:27:21 totally 10:27:24 unreadable 10:27:30 With optional VESA support IIRC 10:28:04 vesa would be ideal. vesa LFB, i have no desire to get into the insanity of earlier stuff 10:28:14 Mostly assembly code though 10:28:36 Yeah, it's VESA1 modes using an LFB. 256 (8-bit) color only 10:28:39 But it works 10:28:50 any plans to add an asm wordlist so there are mnemonics instead of raw hex code? 10:29:03 I440r_: I was having trouble getting it to work. 10:29:06 I'm working on that 10:29:20 But the assembler stuff is going slowly 10:29:25 Then I looked at number.f and realized that . really doesn't need double-cell math. 10:29:28 x86 asm is horrible 10:29:32 cool. since i'm kind of a newbie in asm, so raw hex hurts my brane 10:29:35 tathi im having fbsd problems 10:29:38 :-) 10:29:59 the very first syscall in the extended kernel here barfs with e nomem 10:30:03 The assembler takes syntax like: 100 eax i/r mov, 10:30:03 I440r_: just because the representation is bad doesn't mean the content is. I know its formatted horrible (the HTML one is better), but you can find stuff and use it :) 10:30:05 the brk 10:30:16 i/r? 10:30:18 the html one STILL sucks 10:30:23 crc: why the i/r? 10:30:23 immediate -> register 10:30:28 and yes. ive used that list since before you were born :P 10:30:31 I440r_: in appearance, not in content. 10:30:38 I440r_: hehehe, how old are ya? :) 10:30:46 how about reg? 10:30:48 39 10:30:57 r/r 10:30:57 you've used this list since you were 22? :) 10:30:58 r/m 10:31:00 m/r 10:31:04 younger 10:31:05 ah 10:31:11 the list is that old? 10:31:11 lol 10:31:19 .... 10:31:20 dude 10:31:20 its probably older than 22 eyars 10:31:23 you're as old as my dad 10:31:24 lol 10:31:25 That saves a lot of trickery later on 10:31:41 i trust i can always find someone older than me on #forth 10:31:50 yep. 10:31:51 eerie thing is, chuck moore looks just like my dad 10:31:57 hehe 10:31:57 and my name's chuck 10:32:02 chuck is the oldest of us all 10:32:02 :) 10:32:04 err.. 10:32:23 and you live relatively close to him too 10:32:36 maybe he.. 10:32:49 no, his son lives in san luis obispo 10:33:01 which is a decent drive north of where i am 10:33:16 stalk him 10:33:27 OH YES!!!!! 10:33:39 "who the hell are you?" 10:33:41 LETS STALK CHUCK MOORES SON SO THAT WE CAN HAVE HIM BE A #FORTH REGULAR! 10:33:44 lol 10:33:48 "well... you're the son of god so i'm talking you" 10:33:49 :p 10:34:01 :P 10:34:08 stalking. 10:34:38 maybe, if you mention his name, and he googles for his name, he'll join! 10:35:12 poing: http://www.retroforth.org/dev/experiments/Assembler.txt has a slightly out of date copy of the current assembler code 10:35:33 RetroForth now runs on SCO OpenServer 5.0 too :-) 10:36:06 eww!!! :) 10:36:09 ^_^ 10:36:11 lol? 10:36:39 for all those hobbyists who prefer sco 10:36:52 my forth will run on a computer with any operating system 10:36:57 as long as you boot from the floppy 10:37:07 which forth is that? 10:37:07 :P 10:37:08 (hahaha) 10:37:22 foxfroth 10:37:29 no, kuF 10:37:29 and yes, i purposely mispelled that 10:37:35 :) 10:37:35 One of the contributors ported it so he could use it on the server at the university he attends 10:37:50 funny thing is how picky colorforth is about hardware ... yet it boots fine on my vaio, which is chock full of propietary stuff 10:37:54 RetroForth/Native will run on some PC's 10:38:01 and my gf's custom built vanilla machine ... no luck 10:38:14 :D 10:38:19 I can't CF on ANY of my boxes 10:38:31 ah yes, i guess i should add, a x86 system running any os :p 10:38:34 i just could not wrap my head around cf 10:38:35 me neither 10:38:40 I can't boot CF on ANY of my boxes 10:38:55 neither. 10:38:55 :) 10:39:03 cant run or boot 10:39:05 I have a non-standard boot sector in RF/Native (needs multitrack reads...) 10:39:16 i have no idea why i cant, i just cant 10:39:23 poing: well, CF isn't bad at all once you get the hang. However, doing anything else than trivial calculations is gonna be hell 10:39:34 poing: his pad is hell, his character emit stuff is hell 10:39:44 poing: but i like the color idea, so im implementing it for f2 :) 10:39:52 i remember someone scoffing at moore's "simple ide driver" that it was too simplistic, and didnt cover border cases, chipset blacklist, etc 10:40:14 and i thought "well if that works, why not use that code, and use specialized code" ... specializing compiler 10:40:35 always wonder why OS kernels never use specializing compilers, they'd be like the first to benefit 10:41:52 I think OSes are too complex anyway 10:42:07 * poing remembers seeing some nice tiny drivers in the forth on sun boot proms 10:42:19 yep 10:42:27 OpenFirmware is cool 10:43:14 i hate object code. is the problem with most low-level stuff, nonrelocatable brittle object code 10:43:59 thing that drove me from c++ wasn't the language, it was the linker. object code again 10:45:07 The complexity of the language is what drove me away from C++ 10:45:31 i dont mind the complexity. the syntax is a giant wart tho 10:45:34 forth drove me away :) 10:45:40 WOW!!!!!!!!!!!!! 10:45:44 kickass!!! 10:45:56 fasm's ELF stuff is flexible as hell! 10:46:02 How so? 10:46:08 i looked at stuff like DREAMS and thought "wow that's really warped. but cool." 10:46:16 never thought of forth as high level til i saw that 10:46:25 DREAMS? 10:46:33 not that i'd use it, i'd use something bent around my own mind's funny shape ;) 10:46:57 DREAMS is an early OO forth. it's ... poetic. and utterly strange. 10:47:53 Ah 10:47:59 lemme see if i can dig up some examples 10:48:30 guess it's not that early 10:49:20 http://www.elilabs.com/~rj/dreams/node48.html#listing1 10:50:19 Interesting 10:51:17 cool 10:51:17 :) 10:51:20 you gotta love a DSL with words like IMAGINE and STUPOR and REALITY 10:51:37 :) 10:52:06 I dislike OO languages, but this is still kind of cool 10:52:24 --- join: mur__ (~mur@mgw2.uiah.fi) joined #forth 10:52:28 --- part: poing left #forth 10:52:33 --- join: poing (~chuck@64-121-15-14.c3-0.sfrn-ubr8.sfrn.ca.cable.rcn.com) joined #forth 10:52:45 oops, ^W is not "erase word" in xchat 10:53:26 dreams is pretty incomprehensible syntax to me, but it sure looks impressive 10:55:20 I agree there 10:59:09 * poing needs a project to get started tho. thinking something graphical, turtle graphics like 11:00:32 just seems to be so much underdocumented deep magic when it comes to working with low level video 11:02:07 The RetroForth vga/vesa driver set the video mode (before loading the kernel), then the kernel would find it and provide 'frame', 'hp', 'vp' 11:02:27 Frame being the start of the framebuffer 11:02:51 hp and vp being the height and width in pixels (hp = width, vp = height) 11:03:06 --- quit: mur_ (Connection timed out) 11:03:09 There were other resources too, but those were the main ones 11:03:36 frame being the actual framebuffer? 11:04:06 'frame @' returns the starting address of the framebuffer 11:04:19 that sounds reasonably easy 11:04:48 You could use 'c@' and 'c!' (not sure exactly when those first appeared) to set/read pixels from it 11:04:48 that's in pre-6.x retroforth? 11:05:03 That's the 5.x and early 6.x releases 11:05:14 and in 7.x it's textmode only, right? 11:05:25 * crc is checking for the most recent Native 6.x release 11:05:26 Right 11:05:49 hm. shouldnt be too hard to glom it back in tho, right? 11:05:52 tehsux 11:05:53 :) 11:06:30 http://www.retroforth.org/download/rf-native.zip 11:06:48 Uses 'NASM' to assemble; Release 6.20, so it's not too old... 11:06:54 guess i'll have to create a whole font display engine unless i can figure out how to switch vid modes post-boot 11:07:12 There's 'emit' and some other primitives written in assembly 11:07:16 vesa makes you drop into real mode for that, doesnt it? 11:07:24 In drivers.asm 11:07:24 No 11:07:35 VESA mode is configured in the boot sector 11:08:15 When you boot into RF, enter 'vesa' to set hp, vp, and frame 11:08:29 VESA support isn't flawless, but it works for me :-) 11:09:04 so native is based on 6.20? 11:09:31 That file is 11:09:43 The 7.x releases all share a common core 11:09:57 With a few minor differences in system interface, etc 11:10:29 RetroForth/Native 6.20 uses the same core as all of the other 6.20 releases 11:11:03 ah 11:11:09 There was a RetroForth/Native 6.30 somewhere; I should see if I can find that... 11:12:46 http://www.retroforth.org/download/native.zip 11:13:07 That's RetroForth/Native 6.30 which is from May 25 11:17:09 hm. qemu doesnt seem to like it 11:17:47 --- join: crc_ (crc@21-pool1.ras11.nynyc-t.alerondial.net) joined #forth 11:17:59 back 11:18:01 * crc_ hates dialup 11:18:16 qemu doesnt seem to like the native image 11:18:39 Do you have Bochs? 11:19:01 * crc_ hasn't tested RF/Native 6.x under Qemu 11:19:24 not yet 11:19:33 --- quit: crc (Read error: 104 (Connection reset by peer)) 11:19:46 --- nick: crc_ -> crc 11:20:06 I used Bochs for testing it during development 11:21:34 got bochs now 11:22:08 should be bochs -f bochsrc.bxrc, right? 11:23:20 I think so 11:23:37 I just got 6.30 to run under qemu :-) 11:23:46 hm 11:23:52 gives me naught but a blank screen 11:23:54 On the default image you have to do '320 200 setres clear' 11:23:55 what options did you sue? 11:24:09 qemu.exe -L . -m 4 -fda raw.bin 11:24:57 typed 320 200 setres clear and it turned the screen all white 11:25:00 can't see what i type 11:25:02 that normal? 11:25:12 At first it is 11:25:26 oh there it is 11:25:43 * crc must have f****d up something 11:26:01 heh, did 640 480 setres clear after that, and now i just have every other scanline blank 11:27:02 It's really picky on the resolution matching what's actually there... 11:27:35 --- join: Sonarman (~matt@adsl-64-169-95-189.dsl.snfc21.pacbell.net) joined #forth 11:27:58 maybe qemu doesn't support anything higher 11:28:18 oh wait it would have to, i've seen it running linux 11:28:22 X and all 11:28:33 It uses whatever mode you set in the bootsector 11:28:38 320x200 is the default 11:28:44 --- quit: poing (orwell.freenode.net irc.freenode.net) 11:28:44 --- quit: I440r_ (orwell.freenode.net irc.freenode.net) 11:28:44 --- quit: Robert (orwell.freenode.net irc.freenode.net) 11:28:44 --- quit: skylan (orwell.freenode.net irc.freenode.net) 11:28:49 --- join: Robert (~snofs@c-bf5a71d5.17-1-64736c10.cust.bredbandsbolaget.se) joined #forth 11:28:51 --- join: I440r_ (~mark4@216-110-82-59.gen.twtelecom.net) joined #forth 11:28:56 --- join: poing_ (~chuck@64-121-15-14.c3-0.sfrn-ubr8.sfrn.ca.cable.rcn.com) joined #forth 11:28:56 640x480 is default with VESA 11:29:44 --- join: skylan (~sjh@vickesh01-4677.tbaytel.net) joined #forth 11:30:17 I440r: hey d00d 11:30:37 I440r: whats with the mmap in init_mem? 11:30:46 I440r: and, why is the fd 0? 11:30:47 ohhh have to set it in the bootsect 11:30:55 (fd 0 is stdin, right? o.O) 11:31:21 Awesome, QEMU supports VESA 11:31:24 :-) 11:31:27 :) 11:31:30 qemu does it all 11:31:41 and it's way fast 11:31:47 whats qemu? 11:31:50 another bochs? 11:32:02 arke: yep, and really fast 11:32:04 Except for displaying text when I use 'type' under 7.2 and 7.3... 11:32:18 type? 11:32:34 oh printing 11:32:56 switches to textmode or something? 11:33:46 7.3 *is* textmode at this point 11:33:46 --- nick: poing_ -> poing 11:34:06 But for some reason 'emit' works fine 11:34:09 So does '.' 11:34:16 But 'type' does NOT work 11:34:21 * crc doesn't understand why 11:35:13 qemu sounds cool 11:35:25 can it run cf? 11:35:50 beats me. one easy way to find out 11:36:09 arke: the mmap is to allocate the return stack 11:38:51 arke: fd is ignored in the mmap 11:39:15 arke: because one of the other parameters has a flag MAP_ANONYMOUS (I think?) which tells it to just allocate memory. 11:39:25 (as opposed to actually mmaping a file). 11:39:59 tathi: do you know why he deosn't just use brk? 11:40:14 ColorForth crashed qemu for windows :-( 11:40:39 Sonarman: he does use brk -- brk's out to 1MB, splits it in half for headers and code/data. 11:40:49 i mean, for the return stack 11:40:53 Sonarman: then mmaps a 4KB chunk for the return stack. 11:41:08 why doesn't he just brk an additional 4K for the return stack? 11:41:36 Well...on PPC anyway, mmap tends to allocate in a different place than the application is loaded. 11:41:56 so if you mmap, then you get a segfault if you overflow the rstack, rather than garbling the end of your heap. 11:42:01 Not sure if that's true on x86 though. 11:42:04 ah, ok 11:42:46 the FD has to be -1 if doing anon 11:43:05 but you push an FD of 0 11:43:17 and it still works :) 11:43:24 i x86 yes 11:43:32 in fbsd it HAS To be a -1 11:44:05 oh ok 11:44:07 and as for why i allocate the return stack.... 11:44:17 because i want it to be a grows down stack 11:44:26 and i dint want it to start overwriting head space 11:44:34 its not currently a grows down mapping 11:46:04 Sonarman, is studying the isforth sources ??? :) 11:46:22 how did you init the data stack? 11:46:47 :-) 11:47:34 is it specified in the ELF header or somethign? 11:48:22