00:00:00 --- log: started forth/04.07.09
00:15:50 --- join: foobar (fooboo@dialup-4.152.81.110.Dial1.Atlanta1.Level3.net) joined #forth
00:24:41 --- nick: SolarFire-| -> Solarfire
00:36:55 --- join: qFox (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
00:55:23 --- quit: poingie (Read error: 110 (Connection timed out))
02:09:41 --- quit: Klaw ()
03:05:40 --- quit: fridge_ (Remote closed the connection)
03:32:43 --- quit: aum ()
03:53:43 <Robert> I440r_: It's OK.... It's not like I NEED it.
03:55:39 <Robert> otpg.ø
03:56:01 <Robert> Argh,..
03:56:07 <Robert> Hamster running on the keyboard
04:22:37 <mur> Øøre
04:23:09 --- part: foobar left #forth
04:33:38 --- join: crc (crc@0-1pool176-19.nas6.philadelphia1.pa.us.da.qwest.net) joined #forth
04:50:02 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth
04:51:28 * crc greets Herkamire
04:53:45 <Herkamire> hi crc :)
05:33:07 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth
05:33:39 * crc greets tathi
05:37:50 <tathi> hi crc
05:38:17 * tathi is having trouble with the GNU assembler (again), grr...
05:38:42 <crc> I don't like the GNU assembler
05:39:03 <tathi> me neither, but I don't know of a viable alternative for PPC.
05:39:31 * tathi supposes he should really stop whining and just write one, but...
05:41:19 <crc> I guess you're stuck then. I'm not aware of any PPC assemblers other than GNU :-(
05:41:48 <tathi> well, there is pasm (http://devnull.owl.de/~frank/pasm_e.html)
05:42:18 <tathi> but I'm not sure it's any better.
05:42:54 <tathi> Actually I don't have issues with the way gas does things.
05:42:59 <tathi> I'm just sick of it not working.
05:44:13 <tathi> Like, for instance... if you have a comment char at the start of a line, followed by whitespace, followed by a word starting with a number...
05:44:34 <tathi> it gives you some error about local labels being required to have positive values.
05:44:45 <crc> That's pretty bad
05:45:09 <tathi> I was just looking at the source, it's starting to make sense why it has all these problems.
05:45:16 <tathi> the parsing code is a complete mess.
05:45:41 <tathi> I bet no-one really understands why it works :)
05:45:46 <tathi> Ah well...
05:47:03 <tathi> so how's retroforth going?
05:49:32 <crc> Pretty good
05:49:50 <crc> I've mostly finished testing the 7.0 codebase
05:50:00 <crc> And the documentation is about half done
05:50:43 <crc> There are about 10 words left to add to my test program :-)
05:51:16 * crc needs to scale back the test program. It's mushroomed to 247MB now.
05:51:51 <tathi> :-)
05:53:56 --- quit: I440r_ (Read error: 110 (Connection timed out))
05:54:01 --- quit: [Forth] (Connection timed out)
05:54:21 <crc> I should port it to PPC Linux one of these days...
05:56:14 --- join: I440r_ (~mark4@216-110-82-1.gen.twtelecom.net) joined #forth
06:00:37 <tathi> whatever
06:00:52 <tathi> :-)
06:05:00 <Robert> Hi
06:07:02 <tathi> Hi Robert
06:09:21 <chandler> gruess gott
06:13:40 --- join: poingie (~chatzilla@64-121-15-14.c3-0.sfrn-ubr8.sfrn.ca.cable.rcn.com) joined #forth
06:15:13 * crc greets poingie
06:15:33 <Robert> Hi poingie 
06:18:36 <chandler> gas is a pile of shit, esp on ppc
06:19:09 <chandler> I know of some ppc assemblers that are embedded into other languages but I suppose that's not much of a help
06:24:24 <tathi> chandler: it recognizes a bunch of characters inside strings.
06:24:58 <tathi> and having '#' as the comment character produces some weird stuff because the C compiler uses it for directives.
06:25:09 <tathi> But other than that, I don't have a problem with it.
06:25:29 <tathi> chandler: Are there other things you don't like about it?
06:25:49 <chandler> on OS X ; is the comment character, unlike on every other platform
06:26:02 <tathi> what?
06:26:07 <chandler> IIRC
06:26:14 <tathi> I thought ; was the standard assembler comment character.
06:26:30 <tathi> at any rate, it has been the comment character for almost every assembler I've ever used.
06:27:11 <chandler> on powerpc linux it's the line terminator
06:27:41 <I440r> tathi i liked devpac for the amiga. anything after the first valid mneumonic was a comment :)
06:27:49 <tathi> I440r: :)
06:27:57 <I440r>  push.l d0   comment comment comment
06:28:04 <chandler> heh
06:28:11 <tathi> yup.  I've used one of those.
06:28:14 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
06:28:48 <tathi> though I think my favorite assembler so far was Eric Isaacson's A86.
06:28:48 <I440r> the only assemblers that ive seen that would allow more than one opcode per source line would be forth assemblers so requiring a ; to start a comment seems silly
06:28:56 <I440r> tathi i registered it
06:29:05 <I440r> i have a386 and d386 :)
06:29:18 <tathi> Oh yeah, that was a great debugger too :)
06:29:19 <I440r> i think THAT is the best damned assembler ever written
06:30:13 <Teratogen> C is a portable assembly language
06:30:14 * Teratogen ducks
06:30:36 <crc> C is (mostly) evil!
06:30:39 <I440r> c is not asm and nothing is portable :P
06:30:44 <chandler> tell me how you return a variable number of values on the stack in C
06:30:52 <tathi> What's the line?  "All the power of BASIC with all the ease of use of assembly language"
06:30:53 <chandler> *bzzt*, times up
06:31:02 <I440r> chandler return a pointer to them :)
06:31:03 <Topaz> pahaha
06:31:07 <chandler> I440r: that's not on the stack :-)
06:31:13 <I440r> it can be
06:31:47 <Teratogen> returning a variable number of arguments is ghey.
06:32:02 <I440r> tathi i now have a static IP at home but the wire has to be physically moved off the internal network to the external one :)
06:32:17 <chandler> Teratogen: dude, what kind of forth programmer are you?
06:32:46 <tathi> I440r: cool.
06:35:00 <I440r> im thinking Teratogen was just joking :P
06:35:04 <I440r> he knew ennuff to duck :)
06:35:18 <chandler> yeah yeah :-)
06:36:17 <Robert> Heh.
06:36:28 <Robert> Teratogen is a notorious #minix troll. ;)
06:36:43 <I440r> lol
06:37:00 <Robert> But he owns some nice hardware.
06:37:06 <Robert> Show them your core memory, Teratogen.
06:37:06 <I440r> please tell me nobody actually runs minix these days lol
06:37:26 <Robert> I440r: I just retired my last Minix computer a month ago.
06:37:33 <I440r> robert lol
06:37:41 <Robert> So now the 2 Minix systems are on the shelf :/
06:37:47 <I440r> well. thats you - that doesnt count, we all know your insane
06:37:57 <Robert> Pffft
06:40:01 <crc> I occasionally run a Mininx HD image under Bochs
07:18:40 --- quit: crc ("http://www.retroforth.org/dev/beta_releases/retroforth-7.beta4.tar.gz")
07:38:36 --- join: warpzero (~warpzero@mi035.dn190.umontana.edu) joined #forth
08:18:29 --- quit: Topaz (Read error: 110 (Connection timed out))
08:18:41 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
08:28:17 --- quit: tathi ("leaving")
08:34:55 --- quit: I440r ("bbl")
08:36:20 --- quit: Topaz (Read error: 110 (Connection timed out))
08:36:44 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
08:53:36 --- quit: Topaz (Read error: 110 (Connection timed out))
08:54:19 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
09:11:59 --- quit: Topaz (Read error: 110 (Connection timed out))
09:12:19 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
10:12:05 --- quit: I440r_ (Read error: 110 (Connection timed out))
10:27:15 --- join: tucknip (~Huck@170.141.109.33) joined #forth
11:19:33 --- join: Serg (~z@193.201.231.126) joined #forth
11:19:49 --- quit: Serg (Client Quit)
11:21:39 --- join: Serg (~z@193.201.231.126) joined #forth
11:22:08 <Serg> link now is unstable ;)
11:22:15 <Robert> Hi Serg 
11:22:22 <Serg> hi
11:23:19 <Serg> my head is like kettle now - lack of slee
11:23:24 <Robert> :(
11:24:47 <Serg> i feel a temptation to code something oldshcool sub -4k
11:25:05 <Serg> maybe, noter ;)
11:26:37 <Serg> sad thing, screen driver is broken in win2k and XP - after mode switch, RU letters go away
11:29:25 <Robert> Serg: Hehe, my Forth that I'm messing with now is 3k ;)
11:29:48 <Serg> nice
11:30:00 <Serg> the one i used for my games is 18k
11:30:32 <Serg> but can build sub-1k .com filezz
11:32:25 --- part: tucknip left #forth
11:32:37 <Robert> :)
11:33:00 <Robert> Well, with all libraries and stuff compiled in, it's over 6k
11:33:03 <Serg> the tinyesy usef00l prog, "u2d", is 546 bytes
11:33:22 <Robert> The base Forth system is always compiled in.
11:33:38 <Serg> endline converter, 1-liner
11:34:12 <Serg> : utod begin key dup 10 = if 13 emit then emit again ;
11:34:14 <Robert> Which is pretty nice, since you get all the features of the compiler.
11:34:29 <Robert> I.e. powerful scripting features in a 4k pong game ;)
11:34:54 <Serg> no, w/ compiler it's 18+ K
11:35:22 <Serg> 546 bytes is proggie w/o compiler, sure
11:36:17 <Serg> i have one of u'r Forths, but i found it unwieldy
11:36:33 <Robert> Which one?
11:36:39 --- join: Klaw (~anonymous@ip68-228-92-218.oc.oc.cox.net) joined #forth
11:37:00 <Serg> tforth ?
11:37:22 <Robert> That's the one I'm working on now... The old version is very buggy. 
11:38:03 <Serg> it not even stumbled over bugs - weird startup sequence made me drop it be4
11:38:21 <Robert> http://robert.zizi.org/pub/projects/tforth/tforth.zip (20kB)
11:38:23 <Robert> Oh?
11:38:26 <Robert> What was wrong?
11:38:57 <Serg> the fact what it loads some .f src on startup, and no _easy_ way to disable this
11:39:12 <Robert> Hehe
11:39:17 <Robert> I know, that was a problem.
11:39:20 <Robert> It's fixed now :)
11:39:25 <Serg> k00l !
11:39:34 <Robert> Just type (if you want it to start up with the word "foo"):
11:39:38 <Robert> ' foo is quit
11:39:54 <Robert> fsave filename.com
11:41:00 <qFox> just 0x1000 bytes left
11:41:02 <qFox> to erase
11:41:17 <Robert> :D
11:41:18 --- quit: Topaz (Connection timed out)
11:41:27 <Robert> You're crazy, qFox... Crazy.
11:41:39 <qFox> nope
11:41:43 <qFox> just a rusty old key and my gsm
11:41:44 <qFox> :p
11:41:51 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
11:41:59 <Serg> i'll try then i'll have cheap inet
11:42:00 --- quit: Serg (Read error: 104 (Connection reset by peer))
11:42:39 --- join: Serg (~z@193.201.231.126) joined #forth
11:42:45 <Serg> re
11:42:56 <Serg> what was the reason of crashout ?
11:43:15 <qFox> peer
11:43:18 <Robert> 20:42:02 -!- Serg [~z@193.201.231.126] has quit [Read error: 104 (Connection reset by peer)]
11:43:56 <Serg> heh, i recieved call on the same cellphone
11:44:34 <qFox> lol
11:44:49 <qFox> still internetting @ 75cents a minute? ;p
11:45:07 <Serg> Robert: have u easy i-face for adding ASM words ?
11:45:23 --- join: tucknip (~Huck@170.141.109.33) joined #forth
11:45:24 <chandler> my cell phone has $10/month flat fee net access
11:45:30 <Robert> Serg: No :(  You have to use NASM.
11:45:32 <Serg> qFox: 45 cents megabyte
11:45:50 <Serg> no-no
11:45:51 <qFox> oh i thought you said per minute the other day
11:46:33 <Serg> can ASM modules be loaded by specific word at runtime, or all Forth gotta be recompiled ?
11:46:39 * Serg is not so hell rich
11:48:30 <qFox> yay. done.
11:48:33 <qFox> jezus
11:48:39 <qFox> you need a clear word, seriously
11:48:40 <qFox> :)
11:49:17 <Serg> i never coded in Forth long ago ;( mostly in Perl and PHP
11:49:17 <Serg> qFox: clear word ?
11:49:49 <qFox> Serg> in his other forth is an editor, pretty neat stuff, but core.f is kinda big, and there's no fast way to clear the file
11:50:08 <qFox> had to keep del pressed for about...30 minutes
11:50:16 <Serg> ha-ha !
11:51:15 <Serg> once erasing some multi-file thing from stupid DOS commander, we pressed Y by hammer
11:51:58 <Serg> but the fing plugged together at table w/o box, was beeping so nasty
11:52:26 <Serg> it had buzzer on motherboard, so it cannot be unplugged
11:53:03 <qFox> lol
11:53:07 <qFox> i put a rusty old key
11:53:11 <qFox> which wasnt heavy enough
11:53:21 <qFox> (laptop key no less!)
11:53:35 <qFox> so i put my cellphone on top of it :p
11:54:27 <Serg> did u foto it ?
11:54:35 <qFox> uh, no
11:55:07 <Serg> u missed to share great fun ^))
11:55:29 <qFox> make mental pictures :p
11:55:35 * Serg is carrying around 5$ half-frame camera ;)))
11:57:12 <Robert> qFox: Clear word?
11:57:16 <qFox> never. mind.
11:57:21 <Robert> Serg: Forth has to be recompiled.
11:57:22 <Serg> and 20$ exponometer - matches w/ digicam from dark undertable to sunshine, despite 30y old and wrong batteries
11:57:30 <Robert> qFox: You can clear the editor's buffer with the word "new".
11:58:04 <qFox> and now i hate you
11:58:14 <qFox> 30 minutes of that cellphones life he'll NEVER get back
11:58:21 <Robert> qFox: I told you about it earlier, remember?
11:58:31 <Robert> You never listen to me :(
11:58:34 <qFox> when i did new, i got the core pasted in it, remember?
11:58:42 <qFox> appearantly, the whole core, not just the first line.
11:58:48 <Robert> Then you did something wrong. :)
11:58:53 --- quit: Topaz (Connection timed out)
11:58:57 <qFox> says he who coded it... i only pressed enter :\
11:59:09 <Robert> Tss...
11:59:21 <Robert> That whole system is a quick hack to please the teachers.
11:59:27 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
11:59:46 <qFox> hacks are evil
12:00:04 <Serg> Robert: why not normal screen ed w/ menu ?
12:00:20 <Serg> good old gap based ;))
12:01:42 <Robert> Uhm, what? :)
12:02:00 <Serg> gap based editor ?
12:02:05 <Robert> What is gap?
12:02:19 <Serg> see, u have RAM buff
12:02:28 <qFox> a big cloths selling store chain
12:02:30 <qFox> :p
12:02:37 <Robert> Heh
12:02:43 <Serg> text be4 cursor is in the start of buff, after - at end
12:03:02 <Serg> between is the gap, into what u print
12:03:31 <Serg> then u move cursor, letters are moved across the gap
12:03:42 <qFox> blank
12:03:43 <qFox> space
12:04:05 <Robert> Hmm...
12:04:15 <Robert> I don't think I've used any of those :)
12:04:53 <Serg> this is internal arch ;)) the gap is not on-screen, obviously
12:05:25 <qFox> robert> var tmpd0 400 malloc -> tmpd0 ,will work right?
12:05:47 <Robert> 400 malloc var tmpd0
12:05:51 <Robert> That would be a cleaner way
12:06:02 <Robert> var ( init-value -- )
12:06:14 <Robert> So it needs a parameter
12:06:24 <qFox> oh
12:06:28 <qFox> ohye value
12:06:30 <qFox> blah.
12:06:46 <Robert> Tss...
12:07:08 <Serg> blah.. bleh.. bletch !!
12:07:10 <qFox> ^^
12:08:50 <Serg> how close the languages are !
12:09:29 <Serg> ENG bletcherous - RU bleevat' (to vomit)
12:14:40 <Robert> Heh
12:14:58 <Robert> In Swedish it's kräkas/spy
12:15:32 <qFox> overgeven, kotsen
12:17:19 --- quit: Topaz (Read error: 110 (Connection timed out))
12:17:22 <Robert> Yum.
12:17:28 <Robert> Hmm...
12:17:31 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
12:17:47 <Robert> I just remembered lying on bed with a 486 laptop and a Forth tutorial... :)
12:22:18 --- join: I440r (~mark4@168-215-246-243.gen.twtelecom.net) joined #forth
12:23:24 <Robert> Hi I440r 
12:26:10 <slava> hey
12:26:18 <Robert> And hi slava 
12:26:31 <Robert> How is the debugging doing?
12:27:15 <slava> of the garbage collector? its going well. i've been constructing various heaps to test it a few functions at a time.
12:27:33 <slava> the remaining problem seems to be that some pointers are not updated for some reason (objects get moved as part of collection)
12:28:00 <I440r> hi
12:28:51 --- quit: Serg ()
12:35:20 --- quit: Topaz (Read error: 110 (Connection timed out))
12:35:38 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
12:43:55 --- quit: tucknip (Read error: 110 (Connection timed out))
12:53:37 --- quit: Topaz (Read error: 110 (Connection timed out))
12:54:14 --- join: Topaz (~top@spc1-horn1-6-0-cust217.cosh.broadband.ntl.com) joined #forth
13:39:13 --- quit: I440r (Read error: 54 (Connection reset by peer))
14:00:24 --- join: SDO (~SDO@68.170.20.223) joined #forth
14:06:07 --- join: I440r (~mark4@168-215-246-243.gen.twtelecom.net) joined #forth
14:06:23 <I440r> who is jdoctor ?
14:06:25 <I440r> someone new ?
14:09:33 <qFox> i recognize him from #ai
14:10:19 <I440r> ok
14:10:20 <I440r> :)
14:16:28 --- quit: jDoctor ("leaving")
14:19:01 <I440r> well i wasnt trying to drive him away! lol
14:21:20 <qFox> hehe
14:21:26 <qFox> i used to come to #ai anyways
14:21:38 <qFox> didnt see him there just now, or when i checked yesterday when i noticed him here
14:22:41 --- join: crc (crc@0-1pool176-36.nas6.philadelphia1.pa.us.da.qwest.net) joined #forth
14:25:18 <I440r> wheres tathi :)
14:25:55 <qFox> !seen anyone?
14:26:06 <qFox> ohwell, check logs :p
14:26:42 <I440r> lol
14:26:55 <I440r> pity clog isnt a little more active arround here he?? lol
14:27:08 <crc> tathi	Quit with message ("leaving").	16:28:19
14:27:12 <I440r> i dont think thats why clog exists however
14:28:07 <I440r> ok, now i need a domain name
14:28:19 <I440r> i got me a fast pipe and a static ip
14:28:23 <I440r> what name can i give it
14:29:23 <I440r> i could call it forth.net ??
14:30:33 <crc> Already taken
14:31:02 <I440r> i was just emerging whois hehe
14:31:15 <crc> :-)
14:31:24 <crc> emerge = gentoo?
14:33:13 <I440r> yup
14:33:39 <I440r> what does it mean when a domainname is locked ?
14:33:41 <crc> Is gentoo good? It's one of the few Linux distros I haven't tried
14:33:51 <slava> nah it sucks :-P
14:33:53 <I440r> i think its second to none
14:34:11 <I440r> if you try it do a stage 1 install following the manual
14:35:01 <I440r> the "handbook" they call it
14:35:38 <crc> Domain locking prevents your domain name from being hijacked or accidently transferred. If a transfer to another registrar or hosting company is requested for a locked domain name the transfer request will automatically fail. If you wish to transfer a locked domain name you simply have to login to your control panel and unlock the domain name. 
14:37:36 <I440r> well this is just a domain name someone is sitting on. isnt that illegal now ?
14:37:43 <crc> Nope
14:37:55 <crc> Unless you can prove you own the name as a trademark
14:37:56 <I440r> its the domainname i registered years ago and had to let lapse
14:38:06 <crc> So?
14:38:19 <I440r> no big deal, i just didnt think it was allowed
14:38:30 <I440r> i dont know if i want it back or not :)
14:38:49 <I440r> and if i do it just means i get it through THEM :P
14:39:28 <crc> :-)
14:42:16 * crc likes Google Ads
14:45:03 --- quit: I440r (Remote closed the connection)
14:51:36 --- quit: madgarden (Read error: 110 (Connection timed out))
14:57:56 <qFox> a word is in a dictionary thats in a vocabulary, right?
14:59:17 <crc> I tend to think of a dictionary as containg vocabularies
14:59:22 <crc> Words are defined in vocabularies
14:59:41 <crc> Vocabularies are contained in the dictionary
14:59:49 <crc> At least in RetroForth
15:00:17 <qFox> whats the view on this from the "classic forth perspective"? as it was designed...
15:00:26 <slava> originally there was only one vocabulary
15:00:33 <slava> some forths are still like this i think
15:00:52 <slava> some have two vocabularies, one of immediate words and one of normal words (instead of the immediate flag)
15:00:54 <qFox> the way i see it, its vocabularies, with a search order, that contain a chain of dictionaries, that each contain a chain of words
15:01:10 <slava> on the other hand, some have vocabularies as a means of modularizing code
15:02:18 * crc doesn't like more than two vocabularies: they make the code more complex than it needs to be
15:02:38 <slava> i have a whole bunch :)
15:02:50 <slava> i prefer it to having to think of silly prefixes to distinglish word names
15:04:06 <crc> I write programs, which I can load from blocks or files, and reset the forth system to its initial state before running new ones
15:04:15 <crc> You don't need a lot of vocabularies
15:05:02 <slava> i like to have everything loaded at once
15:05:12 <slava> and dump/restore a memory image between sessions :-P
15:05:49 <crc> That method is too complex for me
15:06:11 <slava> seems simple enough.
15:06:20 <slava> you just need two things in the language - vocabularies, and redefining existing words.
15:06:27 <crc> RetroForth requires hacking several assembly routines to add another vocabulary
15:06:50 <slava> what are the vocabularies?
15:06:51 <crc> 2 Vocabularies: FORTH for callable words & MACRO for compiler macros (immediate words)
15:07:28 <slava> aha nice
15:07:30 <slava> like pygmy
15:08:15 <crc> Yup
15:08:15 <crc> Pygmy is a good Forth
15:08:22 <crc> One of the Forths that inspired RetroForth's design!
15:08:51 <crc> The others being ColorForth, eForth, cmForth, and Herkforth
15:09:20 <qFox> i cant even get colorforth to start :(
15:09:25 <qFox> it just wont budge
15:09:35 <slava> glcolorforth runs on fbsd
15:09:38 <slava> but i odn't know how to use it
15:09:42 <slava> it has an odd keyboard layout
15:09:44 --- quit: Solarfire ("bye")
15:09:54 <crc> I don't use the ColorForth key layout
15:10:03 <crc> qFox: What OS do you use?
15:10:11 <qFox> no i mean bootable
15:10:24 <qFox> oh well under xp it just starts a deamon it seems, with no interface
15:11:03 <qFox> and when i boot nothing happens on this computer, and on my laptop it freaks
15:11:11 <qFox> odd behaviour, but useless
15:12:15 <crc> ColorForth is VERY picky on the hardware it supports
15:12:33 <crc> I've never managed to boot it either :-)
15:13:03 <qFox> ohwell, doesnt matter
15:14:31 <crc> How's the work with ciForth going?
15:15:11 <qFox> i quit ciforth
15:15:37 <qFox> after not being able to do reach other segments i needed, i had enough of it 
15:15:59 <qFox> now attempting to code a forth on one of Robert's forth's
15:16:10 <qFox> forths
15:18:25 --- join: crc_ (crc@0-1pool88-72.nas48.philadelphia1.pa.us.da.qwest.net) joined #forth
15:19:49 <Robert> Which is quite a challenge, since it's terribly buggy. :)
15:20:04 <qFox> yep, i'm driving robert up the wall all the time. its fun
15:20:32 <crc_> Which one, F?
15:20:37 <qFox> yep
15:21:06 <crc_> :-)
15:21:18 <crc_> I like RetroForth/Native better :-)
15:21:26 * crc_ must be biased or something...
15:21:37 <qFox> hmmmmm shall i implement vocabularies (stack of search order) > dictionaries (linked) > words (linked)
15:21:49 <qFox> or shall i leave the dictionaries out of it
15:22:05 <qFox> crc> i'd prefer the native, but i simply cant do it myself
15:22:14 <crc_> Do what?
15:22:19 <qFox> lack the knowledge to build a bootable forth
15:22:26 <qFox> on x86 anyways
15:22:58 <crc_> I'm working on updating RetroForth/Native to the 7.0 codebase
15:23:38 <crc_> I have a new boot sector + A20 initialization code that should eliminate most of the boot problems people faced in the past
15:24:24 <crc_> It's really nice to have a Forth with a single core codebase that can be used under Linux, FreeBSD, BeOS, Windows, and standalone :-)
15:24:34 <qFox> well i'm going to try and build my forth on top of F, leaving F in the dark once my own forth initializes
15:24:36 --- quit: crc (Read error: 104 (Connection reset by peer))
15:24:50 --- nick: crc_ -> crc
15:24:57 <crc> I see
15:25:17 <qFox> not entirely sure at this point how to actually accomplish this, i have idea's, and working on code as we speak, but whether it'll work...
15:25:19 <qFox> time will tell :)
15:27:45 <qFox> i think i'll leave the dictionaries out. they serve no direct purpose
15:43:57 <crc> I use the terms dictionaries & vocabularies interchangeably
15:44:36 <qFox> i was once told that vocabularies existed of dictionaries, so i try to stay by that definition
15:54:30 <crc> In my view, you have a dictionary, which contains pointers to each vocabulary
15:54:41 <crc> It's pretty simple in actual practice
15:55:01 <crc> Only having two vocabularies makes some things (like search order) simple
15:55:11 <qFox> well, i'll have a stack with id's, that serves as the search order
15:55:42 <slava> my interpreter can now save 'images'
15:55:46 <crc> I just search:
15:55:47 <slava> before it could only load them
15:55:48 <crc>  1) FORTH if interpreting
15:55:57 <qFox> there are some things from ciforth i really liked, this slightly different from what ciforth had, but i like it. 
15:55:58 <crc> 2) MACRO, then FORTH if compiling
15:56:20 <crc> slava: can an image be run directly?
15:56:46 --- join: aum (~aum@port-204-54-210.fastadsl.net.nz) joined #forth
15:56:53 <slava> crc, no.
15:57:07 <slava> crc, the interpreter loads it into memory and it becomes the heap
15:57:12 <crc> Ok
15:57:16 <slava> crc, its like a dictionary, with a 'here' pointer and allot etc
15:57:29 <slava> crc, except there is a garbage collector so you never have to explicitly free anything, you just allot it :)
15:57:41 <crc> I never free anything anyway
15:57:56 <crc> (Unless I reset the forth environment or stack)
15:58:03 <slava> well i can have strings that grow to any size 'for free', no fixed buffers
15:58:09 <slava> this is the main advantage IMHO
15:58:49 <qFox> well that seems to be the thing about forth, memmory is not recycled
15:58:57 --- join: LOOP-HOG (~jdamisch@sub22-119.member.dsl-only.net) joined #forth
15:59:22 <LOOP-HOG> hi
15:59:22 <crc> I don't think that slava is writing a real forth, just a forth-like language :-)
15:59:29 <crc> Hello LOOP-HOG
15:59:34 <slava> at the moment my image is 65kb. this is enough code to present an interpreter prompt
16:00:00 <slava> the kernel is a 15kb executable
16:00:29 <LOOP-HOG> maybe 15kb is not so bad, atleast its not a java
16:00:58 <LOOP-HOG> lol
16:00:59 <slava> the kernel is in C
16:01:02 <crc> RetroForth's kernel is 3,881 bytes; including ELF header and ASCII bootstrap code
16:01:26 <crc> The entire memory footprint is just 48,384 bytes
16:01:36 <crc> 15k is too big IMO
16:01:45 <slava> why?
16:01:50 <slava> it provides more than a forth kernel
16:01:51 <LOOP-HOG> I forgot to mention the new urls for HOG-SPACE that are hanging off of the retroforth site
16:02:01 <crc> I know that
16:02:16 <crc> But GC, etc are not essential for most tasks
16:02:22 <crc> Keep things simple
16:02:33 <slava> i think gc is essential
16:02:38 <slava> i make extensive use of linked lists etc
16:02:44 <slava> and its nice that the interpreter never corrupts memory
16:02:54 <crc> What's the performance like?
16:02:55 <slava> since all memory access is through kernel primitives for arrays etc that do bounds checks
16:03:03 <slava> its slower than forth
16:03:10 <crc> How fast?
16:03:11 <slava> and much slower than java
16:03:21 <crc> Slower than Java? That's bad
16:03:31 <slava> java is faster than forth on numerical benchmarks.
16:03:40 <slava> of course the interpreter loads instantly (unlike java)
16:03:41 <LOOP-HOG> slava is free to create any sort of Forth that he wishes
16:03:54 <slava> but if you do, eg, recursive fib, its about 20x slower than java and 5x slower than gforth.
16:04:26 <slava> if i add a native code compiler it will be much faster, faster than dtc forth and comparable to java
16:05:06 <slava> its designed for writing network applications
16:05:18 <LOOP-HOG> i'm hoping to have HOG-FORTH up and running by the end of 2008!
16:05:23 <slava> so you don't have to worry about memory corruption, string overflows, and similar problems.
16:05:40 <slava> in a web app, you spend most of the time waiting for network I/O anyway.
16:06:33 <slava> and pauses of 1/10 a sec or so are ok for garbage collection.
16:08:02 <Robert> Hi LOOP-HOG 
16:08:04 <LOOP-HOG> someday i would like to play with another FOS besides colorForth
16:08:08 <LOOP-HOG> :)
16:08:18 <Robert> I wish mine was cleaner.
16:08:44 <aum> hi - anyone seen I440r?
16:08:47 * Robert likes his Forths small.
16:08:55 <Robert> aum: He was here an hour ago.
16:09:09 <slava> one day i might rewrite factor to meta-compile itself instead of writing the kernel in C
16:09:11 <LOOP-HOG> can anybody recommend a good archiving program that works well from the command line and in windows
16:09:48 <Robert> iirc, GNU tar/gzip/bzip2 are available for Windows.
16:09:51 <aum> LOOP-HOG: easiest option is to get the free winzip command line utils, www.winzip.com
16:10:32 * aum whinges at the rudeness of many people who create zip files - who don't put all the files under a single directory first
16:10:36 <LOOP-HOG> im my experience winzip did not compress as well as BitZipper
16:10:53 <LOOP-HOG> i try to remember to do that aum
16:10:58 <Topaz> there's a prog called '7-Zip' which does rather well
16:11:04 <LOOP-HOG> its a little tricky if you havn't done it b4
16:11:11 <LOOP-HOG> i made that miskake a few times
16:11:13 <Topaz> infact, its internal algorithm beats rar, i believe
16:11:45 <Topaz> aum: heh, we had a rather moronic lecturer tell us once that (for a coursework), we'd lose marks if we put them all in a subdir
16:11:47 <LOOP-HOG> i was thinking of tossing the stupid zip files and creating my own compression format
16:12:15 <LOOP-HOG> that way in order to get my site on your desktop, you would have to download an ANS compliant program
16:12:26 <LOOP-HOG> and then run the .yuk files or whatever through it
16:12:46 <Robert> I think bzip2 does a good job at compressing.
16:12:51 <Robert> It's a bit slow though.
16:12:54 <LOOP-HOG> an ANS comliant Forth source and compile it on your machine and then runing the yuk files thru it
16:13:11 <LOOP-HOG> kind of involved for a newbie though
16:13:20 * qFox hit a few walls
16:13:25 <qFox> concrete walls no less
16:13:27 <Robert> qFox: :D
16:13:29 <LOOP-HOG> probably i'll just be lazy and do what i'm already doing until i decide that they old way was too much work
16:13:48 <qFox> i'm gonna go to bed and think about how to desolve those walls
16:13:51 <qFox> bah :(
16:13:53 <qFox> nite
16:13:59 <LOOP-HOG> bye
16:14:09 --- quit: qFox ("meh ircii's netsplit detection appearantly sux ^^")
16:14:30 <Robert> Hehe
16:14:44 <Robert> I feel proud when my code is driving a dutchman to madness.
16:15:02 <aum> any isforth users here?
16:15:06 <Robert> Yes, I am.
16:15:11 <crc> I use isForth sometimes...
16:15:29 <aum> how do i go about removing the curses stuff in isforth, and using plain stdin/stdout?
16:15:33 <crc> Mostly for as a base for the RetroForth Benchmark
16:16:54 <LOOP-HOG> so many Forths, so little time :^)
16:16:59 <Robert> Hehe.
16:17:55 <Robert> aum: OK, found it.
16:17:59 <Robert> Open up isforth.f
16:18:14 <Robert> fload src/ext/hello.f           \ - nice signon message
16:18:14 <Robert> fload src/ext/status.f          \ - status line display
16:18:19 <Robert> Comment out those lines.
16:18:23 --- quit: warpzero ("Tried to warn you about Chino and Daddy Gee, but I can't seem to get to you through the U.S. Mail.")
16:18:58 <Robert> I guess there are some other things you can comment out there.
16:19:18 --- join: kc5tja (~kc5tja@66-74-218-202.san.rr.com) joined #forth
16:19:20 * Robert starts a commenting spree.
16:19:21 <Robert> Hi kc5tja 
16:19:26 --- mode: ChanServ set +o kc5tja
16:19:38 <kc5tja> I just received another donation for the Kestrel.
16:19:56 <LOOP-HOG> ausome
16:19:56 * crc is going to take a nap before work. See you all tomorrow1
16:19:58 <Robert> Cool. :)
16:19:58 <LOOP-HOG> how $$?
16:20:02 <Robert> Night crc 
16:20:03 <kc5tja> Which is interesting, because I was going to sit down and do some more coding on the emulator and the ROM image for it.
16:20:10 --- quit: crc ("http://www.retroforth.org/dev/beta_releases/retroforth-7.beta4.tar.gz")
16:20:17 <slava> kc5tja!
16:20:19 <kc5tja> LOOP-HOG: $25, as published on my site.
16:20:39 <LOOP-HOG> how much so far?
16:20:58 <kc5tja> This brings the total contribution to $125 from a number of people, none of whom I recognize here on this IRC channel!
16:21:01 <LOOP-HOG> HOG-SPACE has minor update with newer zips
16:21:15 <aum> Robert: works fine, brilliant, hasta la vista curses - now, how do i get rid of the 'au Revoir' message?
16:21:19 <LOOP-HOG> that gives me an idea
16:21:24 <kc5tja> (Or maybe they are here, but they just don't have matching e-mail addresses)
16:21:28 <LOOP-HOG> an order of mag less that 125 is 12.50
16:21:34 <Robert> aum: I suppose that would be in there somewhere. grep is your friend, you know.
16:21:51 <LOOP-HOG> that would be enough for a new URL
16:22:16 <LOOP-HOG> if i expanded HS to 5x its current size and filled in all of the descriptions, and put up a paypal site then maybe somebody would pay for 
16:22:21 <LOOP-HOG> www.hog-space.com
16:22:40 <Robert> HS ?
16:22:48 <LOOP-HOG> HOG-SPACE
16:22:52 <Robert> Oh.
16:22:56 <LOOP-HOG> ga
16:23:15 <kc5tja> Folks, I've been really thinking about computer architcture of late.
16:23:24 <kc5tja> The first generation Kestrel is indeed going to remain a CPU-bound design.
16:23:37 <kc5tja> Everything will be, within reason of course, serviced directly by the CPU.
16:23:46 <kc5tja> It keeps the design of the system cheap, and easy to understand for all.
16:24:05 <kc5tja> But the next step up will introduce expansion slots, I think.  The problem is, how do I manage those slots in the best possible manner?
16:24:25 <kc5tja> I'm thinking of walking away from the traditional I/O architectures found in current computers.
16:24:43 <LOOP-HOG> its your pup dude
16:24:45 <slava> why slots?
16:24:54 <kc5tja> Instead, I'm thinking of binding every two slots to a single I/O coprocessor.  It is the I/O coprocessor which has direct access to memory, not the device itself.
16:24:54 <LOOP-HOG> i would very much like to use a faxmodem
16:25:17 <Topaz> how will you arbitrate memory access?
16:25:19 <kc5tja> This way, bandwidth allocation to each device is entirely programmable.
16:25:22 <LOOP-HOG> and use the kestrel as an overgrown answering machine/fax machine/vending machine/etc
16:25:27 <kc5tja> Topaz: You program it.  Explicitly.
16:25:55 <kc5tja> The same way the IBM System/360 mainframe does it, S/370, S/390, Series/7, AS/400, and more.
16:26:56 <kc5tja> So, for example, a video card would have programs for the I/O coprocessor it's bound to designed to stuff the dot shifter registers under its software control.
16:26:58 <Topaz> do they rely on CPU cache to keep the CPU doing stuff whilst the peripherals are poking the RAM? or is the RAM accessible as independent blocks?
16:27:30 <kc5tja> When you say "they," what do you refer to?
16:27:36 <aum> another isforth question - when i pipe commands to isforth, how do i stop these from being echoed?
16:27:38 <Topaz> well, 'the designers'
16:27:58 <kc5tja> Topaz: Too vague.  I can't answer your question.
16:28:06 <Topaz> just surely a problem with having multiple processors talking to the same RAM is that they can only access it one-at-a-time
16:28:08 <slava> what's a good way to code a integer -> string routine?
16:28:38 <Robert> aum: Don't know about that, sorry.
16:28:41 <kc5tja> Topaz: OK, I see what you're saying.
16:29:07 <aum> wtf??? gtop shows isforth as eating 22MB memory?!?!?
16:29:15 <kc5tja> Topaz: A caching system is pretty much a bit of a requirement anyway; SDRAM suffers huge performance hits if you randomly access data anyway.
16:29:21 <Robert> aum: It's just allocating a little bit of heap space. :)
16:29:32 <Topaz> ok, sure
16:29:33 <Robert> aum: So you won't run out of mem in a while.
16:29:46 <slava> aum, gtop shows my interpreter eating 256mb because it allocates a lot of virtual address space
16:29:50 <kc5tja> Topaz: But this is all future material.  I'm just drawing out a road-map of sorts.
16:29:57 <aum> Robert: shame it doesn't alloc heap space as needed
16:30:06 * aum is thinking of isforth for cgi
16:30:21 <Robert> aum: I tried that once.
16:30:27 <aum> Robert: how'd it go?
16:30:28 <Topaz> so what tasks would a I/O coprocessor actually do?
16:30:37 <Robert> aum: Had some trouble with the I/O...
16:30:49 <aum> Robert: what kinda trouble?
16:30:52 <Topaz> would you make them do things like filesystem management?
16:30:54 <kc5tja> I figure putting the IDE drives on their own I/O coprocessor would eliminate the need for me to support very costly IDE-specific DMA circuitry.  Instead, I can use a plain-vanilla TTA coprocessor to talk to the IDE bus as if it were polled I/O, independently of the CPU.  Then I can amortize the cost of the design by re-using it for other devices in the system.
16:31:25 <Topaz> ah
16:31:29 <LOOP-HOG> Kewl
16:31:36 --- quit: SDO (Connection timed out)
16:31:42 <kc5tja> Topaz: No, although it would theoretically be possible.  I would use the I/O coprocessors for actual data transfer, bit-banging control signals to eliminate dedicated controllers, etc.
16:31:46 <aum> forth could totally rule for cgi, given its speed/compactness
16:32:13 <Robert> aum: So low speed is good for CGI? ;)
16:32:35 <LOOP-HOG> i would LOVE Forth CGI in HOG-SPACE
16:32:57 <slava> i just coded this but didn't test it :)
16:32:58 <slava> : fixnum% ( num -- )
16:32:58 <slava>     base /mod swap dup 0 > [ fixnum% ] [ drop ] ifte >digit % ;
16:32:58 <slava> : fixnum>str ( num -- str ) <% fixnum% %> ;
16:33:18 <kc5tja> Topaz: The easiest way to think of those I/O processors is a really, really, really fast PIC chip with very high bandwidth access to system RAM.
16:33:21 <kc5tja> :)
16:33:30 <kc5tja> Of course, since I'm not making a real mainframe, my RAM bandwidth will obviously be limited.
16:33:33 <kc5tja> But still.
16:33:47 <kc5tja> The concept is true, it has lots of prior art, and it would definitely be something totally unique.
16:34:38 <Topaz> i guess it would be possible to have separate buses for each RAM chip, and bus-switching-things on each processor
16:34:40 <kc5tja> The Amiga comes the closest, with its 25+ DMA channels to various functions.  But they're static channels -- they can do only one thing.  
16:34:46 <Topaz> though that probably is going the way of mainframes, heh
16:35:14 <kc5tja> Topaz: Actually, yes.  :)  That is the approach the Cray supercomputer uses for performing its vector operations too.
16:35:28 <Topaz> i guess with FPGAs it'd actually be feasible too
16:35:37 <Topaz> only downside being routing the tracks around the board :D
16:35:43 <Topaz> (or around the chip? :)
16:37:00 <kc5tja> Topaz: Well, I'm not going to do that just yet.  Cost of manufacture for this thing is dominated by PC board space.
16:38:07 <kc5tja> But, by my math, an I/O processor solution can afford upwards of around 50.35MB/s throughput using I/O processors on a 25MHz bus.
16:39:20 <kc5tja> Well, let me be more clear.
16:39:36 <kc5tja> 12.6MWps, where a word is a 32-bits.
16:40:05 <kc5tja> So since each word has four 8-bit bytes, that comes out to 50.35MBps.
16:40:24 <kc5tja> (CPU bus runs at 25.175MHz, btw)
16:40:31 <kc5tja> or rather, will.
16:41:31 <kc5tja> Anyway, the advantage of this is simple: it lets people build *dumb* hardware (read, cheap), and the DMA-like functionalities for the device can be programmed into the integrated I/O processors on the Kestrel-II's motherboard.
16:43:31 <kc5tja> Not sure if that makes any sense or not.  But at least it's an idea, and one I feel which has been neglected throughout history.
16:44:27 <Topaz> yeah, does sound very cunning
16:44:28 <LOOP-HOG> we have control over the hardware ports unlike in XP
16:44:48 <Topaz> and gives ultimate flexibility over protocols and stuff
16:44:55 <LOOP-HOG> its old hat but RS-232 would be welcome
16:45:04 <Topaz> i suppose you could implement all the IO drivers in the coprocessors
16:45:22 <kc5tja> LOOP-HOG: RS-232 can be implemented entirely as I/O controller software -- there would be no distinct UART chip in this system.
16:45:35 <Topaz> so stuff like ethernet cards would be a CS8900A hooked up directly to a coprocessor, with software bashing it and producing ethernet frames
16:46:05 <LOOP-HOG> it would depend if i put it in my build of the kit i buy or not
16:46:07 <kc5tja> Topaz: Right.  The I/O controller would be responsible, once informed a frame of data is complete, for scatter-gathering data and blasting it out its serial port.
16:46:21 <aum> Robert: huh?
16:46:30 <Topaz> yeah, just it'd be possible to design a very simple chip-chip protocol for generic IO handling
16:46:35 <aum> Robert: isforth seems to run at speeds comparable to C
16:46:49 <Topaz> and doing networking and things on the actual processor will be childsplay
16:47:14 <Robert> aum: I think gcc -O3 produces slightly faster code. :)
16:47:17 <kc5tja> Topaz: Yep.  Imagine simple firewalling occuring right on the I/O processor -- no need to burden the main CPU with the simple stuff.  Unless you wanted it to.
16:47:37 <aum> Robert: isforth rules for size though
16:48:00 <Robert> aum: Yep, and speed of compilation.
16:48:23 <kc5tja> And if you really needed it, you could use an I/O processor for an empty slot to do other processing too.  An I/O processor is a terrible thing to waste.  :D
16:48:26 <aum> Robert: so what I/O trouble did you have with isforth/cgi?
16:49:05 <LOOP-HOG> my mind is an analog i/o processor ( said like speak n spell )
16:49:44 <Robert> aum: I don't remember anymore. 
16:54:22 <kc5tja> Anyway, these are just some ideas that I'm kicking around my head.
16:55:22 --- quit: cmeme (Remote closed the connection)
16:55:26 <kc5tja> I still need to finish the basic machine architecture before I consider going on to more advanced operations.
16:56:16 --- join: cmeme (~cmeme@216.184.11.30.unused.swcp.com) joined #forth
16:58:39 <LOOP-HOG> there is a bit of Forth news 
16:59:05 <LOOP-HOG> people are getting serious about putting a complete copy of Thinking Forth online and Leo Brodie gave permission
16:59:11 <LOOP-HOG> also  '336
16:59:16 <kc5tja> YAY!!!!!!
16:59:20 <kc5tja> IT'S ABOUT TIME!!!
16:59:22 * kc5tja dances!!
16:59:33 <LOOP-HOG> :^D
16:59:49 <slava> cool
16:59:52 <slava> starting forth is nice
17:00:12 <LOOP-HOG> =:^D-|-< 
17:00:27 <kc5tja> Yes.  I am looking to maybe use Starting Forth as the skeleton document for my own Forth implementation on the Kestrel.  Of course, I'd have to get permission, but...
17:01:10 <kc5tja> But I've ALWAYS wanted a copy of Thinking Forth so I could read it.  
17:01:20 <kc5tja> I've never, ever been able to find a copy, however.
17:03:07 <kc5tja> I also discovered today that there is a San Diego Forth Interest Group.
17:09:32 <aum> Robert: I see what you mean about isforth giving I/O problems for CGI
17:09:36 <LOOP-HOG> goodie
17:09:38 <Robert> aum: :)
17:09:51 * aum tries piping isforth's output to cat
17:09:52 <LOOP-HOG> we don't really have a portland fig
17:09:54 <Robert> I somehow doubt there is a Stockholm FIG.
17:10:08 <kc5tja> Well, I don't think the SD-FIG is particularly active.
17:10:23 <kc5tja> Of all the user groups on the list that I saw, the FIG was one of four that didn't have a website.  :D
17:12:17 <Robert> Ouch, that's bad.
17:12:28 <kc5tja> He, yeah.
17:12:32 <kc5tja> Heh even
17:12:32 <Robert> The local ham radio club has a web site, but it's VERY inactive.
17:15:24 * Robert just realizes his Forth doesn't support negative numbers.
17:15:37 <kc5tja> See?  You DON'T need signed numbers!!  :D
17:15:54 <aum> ahh, got isforth cgi working
17:16:07 <Robert> 2 neg  <--- that's how -2 should be written. ;)
17:16:12 <Robert> aum: Nice! :)
17:16:36 <kc5tja> Man, I'm really glad to see such increased activity with Forth - if not Forth itself, then heavily Forth-inspired environments.
17:16:38 <slava> what if you need a literal 2? [ 2 neg ] literal is clumsy
17:16:39 <aum> Robert: replaced terminfo floads with a 'faketerm.f', which contains dummy words for white, >bold etc
17:17:20 <kc5tja> slava: But DOABLE.  >:)
17:17:26 <kc5tja> slava: j/k
17:17:43 <LOOP-HOG> maybe somebody with FORTH CGI could sponser a HOG-SPACE style on their server and then I could make a special style that has the CGI stuff in it.
17:17:59 <slava> you could make a parsing word
17:18:01 <slava> neg 2
17:18:09 <slava> would read a number, neg it and , it
17:18:20 <kc5tja> -: 2
17:18:31 <aum> the main problem with forth/cgi is the tendency of forths to start up with a statically-sized heap
17:18:33 <slava> kc5tja, then make - a denotation :)
17:18:41 <Robert> That's a pretty good idea... Thanks.
17:19:28 <kc5tja> slava: I have, after depositing a $666 check in my account, $205 in my account.  Believe me, I have much more important matters to contend with at the moment.
17:24:53 --- quit: Tomasu ("SIGWTF")
17:26:17 <Robert> Heh, speaking of 666, my Forth just reached 6.66 kB with everything compiled into it.
17:26:34 <LOOP-HOG> 666
17:26:38 <kc5tja> Heheh :D
17:26:48 <slava> mine is 66kb :)
17:27:01 <Robert> The ultimate proof - Forth _is_ evil.
17:27:20 <kc5tja> My Forth cross compiler, in GForth, is 50 lines.  Does that count?  ;D
17:27:28 <kc5tja> (not including lines dedicated to comments)
17:27:30 <LOOP-HOG> i should check my program size
17:28:35 <Robert> kc5tja: That's pretty neat. :)
17:28:41 <LOOP-HOG> nah 61344 bytes
17:28:48 <slava> kc5tja, i wish i could fit my source into 64x16 blocks :-P
17:28:53 <LOOP-HOG> you dudes r all pArAnOiD
17:29:25 <kc5tja> slava: My cross assembler is actually in files.
17:29:36 <kc5tja> slava: But yeah, I overwhelmingly prefer blocks to files as a general rule.
17:30:05 <kc5tja> However, the 50 lines of Forth source is for the FTS1001.  It'll naturally be quite a bit bigger for the x86.
17:30:23 <kc5tja> I would predict close to 100 lines for the x86 version (that someday, I'll write).
17:31:37 <Robert> Adding support for "-x" form was the easiest way out...
17:42:38 <LOOP-HOG> i must go
17:42:39 --- quit: LOOP-HOG ()
18:16:48 <slava> my .s almost works ;)
18:19:00 <kc5tja> :0
18:19:04 <kc5tja> man.
18:19:11 <kc5tja> :) <-- that's what I wanted to write.
18:25:43 --- join: I440r (~mark4@216-110-82-1.gen.twtelecom.net) joined #forth
18:29:48 <Robert> Hey I440r... aum has been looking for you
18:30:22 <I440r> aum is dave mcnab right?
18:30:31 <I440r> im emailing him now :)
18:32:11 <Robert> Not sure... Either way he was asking about CGI in IsForth
18:32:58 <kc5tja> re I440r 
18:33:14 <I440r> hi :)
18:33:39 <kc5tja> I440r: You know that the Kestrel software now can print text to the screen, right?  :D  I'm still so proud of that.
18:33:53 * kc5tja will be working on an implementation of TYPE tonight, and maybe start hacking on the keyboard input code.
18:34:56 <chandler> kc5tja! sounds like you've come a long way with Kestrel
18:37:35 --- join: TheBlueWizard (TheBlueWiz@pc5fdn1d.ppp.FCC.NET) joined #forth
18:37:35 --- mode: ChanServ set +o TheBlueWizard
18:41:45 --- quit: kc5tja ("Lost terminal")
18:41:59 --- join: kc5tja (~kc5tja@66-74-218-202.san.rr.com) joined #forth
18:42:04 --- mode: ChanServ set +o kc5tja
18:42:05 <TheBlueWizard> hiya kc5tja
18:47:59 <I440r> tbw :)
18:49:23 <TheBlueWizard> how's life?
18:49:31 <TheBlueWizard> hiya I440r :)
18:50:03 <TheBlueWizard> or to make it more Forth-like: I440r hiya :)
18:50:53 <Robert> WizardBlueThe, hi. :)
18:51:24 <I440r> lol
18:51:34 <TheBlueWizard> heh...Robert hiya
18:56:40 * TheBlueWizard decides this "name" hiya pattern is much better, so he shall greet everyone from now on
18:57:20 <TheBlueWizard> and of course I will localize it as called for...like: mur terve! :)
18:57:34 <Robert> In Swedish it's "hej".
18:57:45 <kc5tja> re TheBlueWizard 
18:57:58 <TheBlueWizard> I know that one :)
18:58:18 <TheBlueWizard> kc5tja hiya...how's life?
18:58:19 <Robert> TheBlueWizard: But you didn't localize my greeting!
18:58:25 <kc5tja> TheBlueWizard: Not too good at the moment.  :/
18:58:31 <kc5tja> Once again, i have to make a partial rent payment for this month.
18:58:43 <TheBlueWizard> Robert hej...better? :)
18:58:48 <Robert> TheBlueWizard: Yes. ;)
18:58:49 <kc5tja> It seems I have a few more months of sustaining myself on Macaroni and Cheese and Ramen.  >:(
18:59:05 <I440r> kc5 :(
18:59:18 <TheBlueWizard> ouch! what's eating into your pocketbook?  Kestrel project?
19:03:20 --- join: Balzac (~balzac33@pcp02022903pcs.rthfrd01.tn.comcast.net) joined #forth
19:04:31 --- part: Balzac left #forth
19:08:04 <kc5tja> No.
19:08:19 <kc5tja> Just catching up on old expenses, and my website bill came in at a most inopportune time.  :/
19:08:54 --- quit: ChanServ (bear.freenode.net irc.freenode.net)
19:08:54 --- quit: ianp (bear.freenode.net irc.freenode.net)
19:08:54 --- quit: slava (bear.freenode.net irc.freenode.net)
19:08:54 --- quit: cmeme (bear.freenode.net irc.freenode.net)
19:08:54 --- quit: lalalim (bear.freenode.net irc.freenode.net)
19:08:54 --- quit: htp123 (bear.freenode.net irc.freenode.net)
19:09:22 <I440r> hmmm we seem to be having alot of those lately
19:09:38 --- join: ChanServ (ChanServ@services.) joined #forth
19:09:38 --- join: cmeme (~cmeme@216.184.11.30.unused.swcp.com) joined #forth
19:09:38 --- join: lalalim (~lalalim@p508AA795.dip.t-dialin.net) joined #forth
19:09:38 --- join: slava (~slava@CPE00096ba44261-CM000e5cdfda14.cpe.net.cable.rogers.com) joined #forth
19:09:38 --- join: ianp (ian@inpuj.net) joined #forth
19:09:38 --- join: htp123 (~tehsux@S0106000d6151238b.gv.shawcable.net) joined #forth
19:09:38 --- mode: irc.freenode.net set +o ChanServ
19:10:15 <TheBlueWizard> ah...like having a car insurance bill coming in at an inopportuner time :-]
19:16:10 <kc5tja> Yes.  :(
19:16:20 <kc5tja> Yes.  :(
19:16:24 <kc5tja> In that order.
19:16:37 * kc5tja really needs to find an improved source of income somehow.
19:18:15 --- quit: Topaz (Remote closed the connection)
19:23:03 <kc5tja> The current job IS letting me (ssslllooowwwlllyyy) catch up on my past debts.
19:23:54 <kc5tja> But it is annoying to suddenly realize you're $400 in the hole because of people who fucking charge their accounts a month and a half after you post a fucking payment.
19:45:34 <aum> what's a forth word that does ( x1 x2 x3 -- x1 x2 x3 x1 x2 ) ?
19:46:05 <I440r> none that i know of
19:46:18 <I440r> >r 2dup r> -rot :)
19:46:42 <aum> ok, thx
19:46:45 <I440r> or 3dup drop
19:46:49 <I440r> if you have 3dup
19:46:56 <TheBlueWizard> call it a double-dutch-leapfrog :)
19:47:25 <kc5tja> How about MOVER, for Mixed-precision OVER.  :)
19:47:44 <I440r> lol
19:58:41 --- join: snowrichard (richard@adsl-068-209-159-248.sip.shv.bellsouth.net) joined #forth
19:59:34 <snowrichard> hello
20:00:13 --- quit: snowrichard (Client Quit)
20:00:14 <TheBlueWizard> snowrichard hiya
20:00:31 <TheBlueWizard> grr
20:03:15 --- join: Tomasu (~moose@68.149.209.223) joined #forth
20:11:05 <TheBlueWizard> gotta go...all bye!
20:11:27 --- part: TheBlueWizard left #forth
20:19:11 --- quit: Teratogen ("SKYKING, SKYKING, DO NOT ANSWER")
20:32:35 --- join: snowrichard (richard@adsl-068-209-159-248.sip.shv.bellsouth.net) joined #forth
20:35:49 <kc5tja> wb snowrichard 
20:36:14 <snowrichard> hi
20:36:20 <snowrichard> busy week.
20:36:21 <slava> hi
20:36:35 <snowrichard> took a spill on sunday and was in hospital a few days
20:38:00 <snowrichard> now I have a black eye and an abrasion on my forehead, but they could find nothing to explain why I passed out.
20:39:07 <aum> any gforth users here?
20:39:21 <snowrichard> I have installed gforth and have not used it yet.
20:39:49 <snowrichard> so not me :)
20:40:23 <aum> weird stuff - gforth is lightning fast, but its dyn lib calls are painfully slow
20:40:41 <Herkamire> gforth is fast?
20:40:48 <aum> yep
20:40:58 <Herkamire> I've used it a bit
20:41:59 <Herkamire> didn't occur to me as fast, but I haven't used a whole lot of other forths to compair it to.
20:43:05 --- quit: snowrichard ("ChatZilla 0.9.52B [Mozilla rv:1.6/1]")
20:43:23 <Herkamire> what's a decent speed benchmark for forth?
20:43:32 <kc5tja> aum: I use gforth extensively, but not in any advanced way.
20:43:58 <Herkamire> kc5tja: what would you consider advanced?
20:44:33 <aum> just felt weird - gforth is extremely fast on a recursive fib test, but its c function calls are way slow
20:45:32 <Herkamire> oh, that' be a nice test. do you have the code for fib laying around?
20:45:44 <Herkamire> I don't feel like writing it again atm
20:46:59 <aum> : fib dup 3 < if drop 1 else 1- dup recurse swap 1- recurse + then ;
20:48:32 <slava> gforth is 6x faster at this than factor :)
20:48:59 <kc5tja> Herkamire: Knowing all the implementation details of gforth.
20:52:45 <kc5tja> YAY!  I just get 'defer, '', and 'is working.
20:53:22 <kc5tja> Folks, I hate to tell you this, but gforth happens to be one of the slower Forth environments for Linux too (although it is one of the fastest of its class).
20:53:45 <slava> eventually factor will compile to native code
20:54:06 <kc5tja> gforth is, on average, 5 times slower than maximally optimized GCC.
20:54:18 <slava> which is fast for an interpreter
20:54:47 <slava> gforth is about 3x slower than java which is not bad
20:55:08 <kc5tja> slava: Well, 3x slower than your JVM.
20:55:18 <slava> when i say 'java' i mean 'sun's jvm'
20:55:25 <slava> which is the only decent one
20:55:25 <kc5tja> Sun's JVM is pretty freaking slow.  :)  I think gforth is noticably faster than the Sun JVM on my particular installation.
20:55:33 <slava> at doing what?
20:55:35 <slava> startup speed?
20:55:42 <kc5tja> No, just general processing.
20:55:52 <slava> how did you determine this?
20:56:04 <kc5tja> Observational evidence.
20:56:11 <kc5tja> I did not do a scientific benchmakr.
20:56:15 <kc5tja> And I apologize if I hurt your pride.
20:56:20 <slava> a lot of java libraries are bloated but this does not mean the VM itself is slow
20:56:35 <kc5tja> slava: What makes the VM slow is its horrid memory management.
20:56:46 <slava> how is it horrid?
20:56:48 <kc5tja> On my box, the JVM thrashes the page memory.
20:56:55 <slava> you have 4mb ram? :)
20:56:55 <kc5tja> It's constantly swapping to disk.
20:57:02 <kc5tja> slava: Do I need to have more?  I shouldn't.
20:57:06 <slava> it uses a lot of memory, yes, but the gc is fast.
20:57:07 <kc5tja> (but to be honest, i have 640MB)
20:57:24 <slava> i never notice gc pauses (but i do notice huge memory usage :)
20:57:33 <kc5tja> GC works *nice* as long as it doesn't think it has infinite amounts of memory (which is why it's slow; when the GC kicks in AND it's paging, ...ZZZzzzzz....)
20:57:40 <slava> right now jedit is using 64mb of ram, which is too much for a text editor
20:58:08 <kc5tja> We have software that uses Java on network servers at work.
20:58:15 <kc5tja> Good lord, they are slow as molasses.
20:58:43 <slava> java for network serving is slow unless a) you run latest jvm b) you tweak your code to hell c) you have a lot of ram.
20:58:48 <slava> consultants make $$$ from this knowledge
20:58:49 <kc5tja> Some 11 to 20 Java instances on a 450MHz processor, even with 512MB, is a good recipe for effectively killing a server.
20:59:40 <kc5tja> slava: Sun's JVM allocates a fixed chunk of RAM for use as its memory arena -- if there is no internal RAM to handle this, it'll page out to disk.
20:59:47 <Herkamire> herkforth is 3.5X faster than gforth at the recursive fib
20:59:57 <slava> Herkamire, nice :)
21:00:16 <slava> kc5tja, all gc languages do this though.
21:00:20 <kc5tja> slava: This is what makes Java so slow.  It's not the bytecode interpreter -- it's the swapping to disk.
21:00:36 <slava> like i said, its an issue of having enough ram
21:00:53 <slava> how do you think sun sells those big servers -- "run your java web apps on these!"
21:01:15 <Herkamire> that's rediculous
21:01:36 <slava> but if a) you have a lot of ram b) you don't use bloated libraries then its reasonably fast
21:01:40 <kc5tja> Look, we're not going to install 20GB of RAM just to run 20 instances of Java for StoreSense, Miva, or whatever shopping cart program the client is using.
21:01:42 <Herkamire> how do you expect a server to handle 20 concurrent requests if each java process takes up 100MB
21:01:46 <slava> my game runs at 100fps (but of course the graphics is rendered in hardware, so...)
21:01:56 <slava> Herkamire, only run 1 java process
21:02:04 <slava> Herkamire, it doesn't spawn a new vm for each request
21:02:15 <kc5tja> slava: Yes, ours does.
21:02:22 <slava> kc5tja, that's just silly
21:02:25 <Herkamire> we're talking about webservers here
21:02:32 <kc5tja> slava: No, that's how *every* network daemon works.
21:02:32 <slava> use something like tomcat or jboss
21:02:38 <kc5tja> slava: WE ARE using Tomcat!!!
21:02:44 <slava> kc5tja, java supports multiplexed i/o and threads.
21:02:52 <slava> there is never a need to spawn a new VM instance to handle a request
21:03:01 <slava> you're using something like inetd?
21:03:03 <kc5tja> slava: Tell that to the authors of StoreSense then.
21:03:16 <slava> i don't take responsibility for badly written apps
21:03:21 <kc5tja> slava: No, it's launched from Apache, of course.
21:03:27 <kc5tja> Did I say you did?
21:03:28 <slava> as a cgi?
21:03:45 <kc5tja> slava: I have no idea.  I just know what I see as a technician who is constantly killing and restarting it.  :)
21:04:05 <slava> well your first mistake is using apache with cgi and not a java httpd
21:04:15 <kc5tja> slava: no, that's not our mistake.
21:04:30 <slava> well, the mistake of whoever set this POS up
21:04:31 <kc5tja> That's a requirement by StoreSense and Miva.
21:05:58 <kc5tja> As for what application I used to compare Java to GForth, though it is none other than Eclipse.
21:06:15 <slava> you're comparing eclipse with what?
21:06:15 <kc5tja> (which is a totally BAD ASS integrated development environment, btw -- HIGHLY recommended)
21:06:20 <slava> no equivalent written in forth exist
21:06:27 <kc5tja> Just the GUI aspect.
21:06:28 <slava> i've tried eclipse, i hate the gui toolkit
21:06:34 <slava> its slow and buggy
21:06:55 <kc5tja> I haven't coded for it, but the facilities that Eclipse uses seem stable enough.
21:07:14 <kc5tja> I've never had Eclipse die.  It's very stable, in my experience.
21:07:32 <slava> did you use the gtk or motif gui?
21:07:39 <kc5tja> GTK
21:07:52 <slava> i found that one very buggy. it was a while ago, though.
21:08:17 <Herkamire> I hope nobody uses the recurive fib for anything but benchmarking. I implemented fib with a loop, and the code is shorter, and executes before you get your hand off the enter key
21:08:37 <slava> kc5tja, what was the forth GUI you compared it with?
21:08:44 <slava> i've never seen a gtk app written in forth but surely its possible.
21:08:56 <kc5tja> The FTS1001 wouldn't be able to execute a recursive fib benchmark; the return stack depth is only 16.  :D
21:09:00 <I440r> : fib 0 1 rot 1 do tuck + loop nip ;
21:09:39 <slava> : fib 0 1 rot [ tuck + ] times ;
21:09:40 <I440r> recursive fib recurses on itself 1 hundred and 13 million times to calculate 40th fib
21:09:46 <kc5tja> slava: I'm talking about general application performance.  Typing a character in Eclipse, there is some delay.  There is none in GForth.  Etc.
21:09:47 <slava> oh that needs a 'nip' too.
21:09:50 <aum> I440r: that's the iterative version - not as useful for benchmark
21:09:50 <Herkamire> fib2 1 swap 1 swap 2- for over + swap next drop ;
21:09:55 <kc5tja> slava: Like I said.  I DID NOT DO A SCIENTIFIC COMPARISON.
21:09:57 <kc5tja> Sheesh.
21:10:16 <I440r> aum its useful for showing how an itterative method beats a recursive one here hehe
21:10:18 <slava> well its a bit silly comparing response time of gtk+ and console...
21:10:24 <kc5tja> No, it's not.
21:10:34 <kc5tja> Many GTK apps are every bit as responsive as a console.
21:10:35 <slava> run a java console app, there's no delay
21:10:37 <kc5tja> On my box, at least.
21:10:41 <slava> no way, gtk is slow-as for me.
21:10:52 <slava> slower than qt.
21:10:55 <kc5tja> slava: I have an 800MHz Athlon, with GTK compiled for my machine.
21:10:57 <kc5tja> It's not slow.
21:11:09 <kc5tja> Qt is **VERY** slow on my box.  I switched away from KDE precisely for that reason.
21:11:21 <slava> it probably depends on X drivers
21:11:23 <aum> I440r: that's an ever-so-slight understatement :)
21:11:24 <kc5tja> And I'm still looking for a suitable replacement for KMail.  As soon as I find one, I'm switching instantly.
21:11:29 <I440r> aum lol
21:11:33 <slava> kc5tja, i use mozilla mail.
21:11:35 <Herkamire> qt has been slow for me. gtk is fast
21:11:43 <kc5tja> No thanks.  I really don't like Mozilla Mail.
21:11:48 <aum> i love recursive fib because it demonstrates how abusive recursion can be
21:11:49 <I440r> recurive method done once... itterative method done 90 million times hehe
21:11:55 <Herkamire> qt is so bloated I've given up on installing it
21:12:02 <I440r> recursive fib is doubly recursive
21:12:04 <aum> Herkamire: wise move
21:12:22 <Herkamire> I started building qt and it just started swapping like crazy after a bit. I let it go for a few hours, then killed it.
21:12:29 <aum> qt is bloat-city
21:12:32 <kc5tja> I currently use either Ion (preferred) or amiwm (very close second, used only when Ion's default key bindings get in my way for an application).
21:12:33 <slava> i've never had this problem
21:12:39 <Herkamire> I never liked anything from kde or qt
21:12:46 <aum> the funniest sight is qt running on a box slower than 400MHz
21:12:50 <slava> i remember i had 32mb of ram in my old computer, and compiling qt took ages
21:13:01 <kc5tja> Herkamire: I like some programs from KDE.  KMail is nice, but it's too sluggish for me.
21:13:05 <slava> nowadays i don't care about 'code bloat' as much :)
21:13:13 --- join: snowrichard (richard@adsl-068-209-159-248.sip.shv.bellsouth.net) joined #forth
21:13:25 <Herkamire> I can't stand sluggish programs.
21:13:31 <Herkamire> I use Mutt. I'm happy with it
21:13:40 <kc5tja> slava: I have an 800MHz Athlon, 640MB of RAM, and I still use and insist on window managers that are less than 256KB in size.  amiwm is less than 80KB in size on my box.  :D
21:13:42 <Herkamire> "all mail clients suck. this one just sucks less"
21:13:43 <slava> i don't like mutt because you can't really view > 1 mail at a time.
21:14:14 <Herkamire> *I* can't read more than one thing at a time
21:14:17 <kc5tja> At work I use twm.  :D
21:14:56 <kc5tja> i want to make a POP client for Forth, but I'm wholesale uncertain how to archive the messages.
21:15:09 <Herkamire> maildir
21:15:30 <kc5tja> Herkamire: Possibly, but it's not good for indexing or searching.
21:16:04 <Herkamire> yeah, my only complaint with mutt, is that it's starting to get slow to load as my mailboxes are getting huge
21:17:17 <kc5tja> the only problem I have with Mutt is its default configuration seems needlessly out of whack with what everyone else uses.
21:17:32 <kc5tja> It seems like everyone who downloads it goes through at least two hours pouring over the default configuration file, tweaking things.
21:17:39 <Herkamire> 1626 messages, takes a few secconds
21:17:48 <kc5tja> And then actually USING it is another matter all-together -- WAY to easy to irretrievably delete messages.
21:17:58 <kc5tja> Herkamire: It's faster than PINE.  :D
21:18:39 <aum> mutt is painful, especially for folder navigation
21:18:44 <kc5tja> Yes.
21:18:50 <Herkamire> yeah, folder nav is lame
21:18:52 <aum> the *nix world still awaits a decent console email client
21:19:05 <Herkamire> it dosn't seem to have the concept of relative paths
21:19:07 <kc5tja> I'm not sure I could improve on it, personally, but hey, since it's Forth, it is imminently hackable.  :D
21:19:22 <aum> a mail client in forth?? which one?
21:19:30 <kc5tja> aum: The one I haven't written yet.
21:19:32 <Herkamire> kc5tja: good point about the configuring
21:19:43 <Herkamire> it did take me quite a while to get it the way I wanted it
21:19:58 <aum> why do so many curses apps totally fail to restore the terminal settings on exit ?!?!
21:20:45 <kc5tja> I think if I were to write a POP client in Forth, I would probably use Citadel BBS-like navigation.
21:21:02 <Herkamire> I have 38 meaningfull lines I put in my .muttrc, pluss the standard PGP stuff
21:21:30 <Herkamire> why do apps have to restore terminal settings???
21:21:33 <Herkamire> it's so lame.
21:21:49 <kc5tja> Indeed.
21:21:53 <Herkamire> every app with anything but a line based text input system has to catch all kinds of signals and crap.
21:22:27 <Herkamire> you have to go through horrible contortions just to get keypresses when they are hit, and I don't even know if it's possible to find out about keyup events under X
21:22:36 <aum> not true
21:22:43 <aum> it's an easy hack to put input into raw mode
21:22:48 <aum> and it doesn't even require curses
21:23:01 <Herkamire> you can get it in raw mode under x?
21:23:20 <aum> of course!
21:23:24 <Herkamire> how?
21:23:34 <aum> gimme a sec, i'll paste the code
21:23:53 <kc5tja> Herkamire: X11 generates both key up and key down events for every key and mouse button.
21:24:16 <kc5tja> A *console* may filter these events out to make it look like a raw ASCII stream.
21:24:22 <kc5tja> But X11's core protocol does have it, and it is used.
21:24:23 <Herkamire> kc5tja: I can get them with SDL, I just wasn't aware you could get them through STDIN under eg xterm
21:24:41 <kc5tja> xterm may or may not support it.
21:24:46 <kc5tja> I'm talking about raw Xlib.
21:24:47 <slava> xterm is an abdomination
21:24:50 <slava> why are we still emulating vt100
21:25:10 <Herkamire> I use rxvt because it's faster and less flickery
21:25:20 <aum> Herkamire: #include <termios.h>
21:25:20 <aum> struct termios newtset, oldtset;
21:25:20 <aum> newtset = oldtset;
21:25:20 <aum> newtset.c_lflag    &= ~(ECHO | ICANON);
21:25:20 <aum> newtset.c_iflag    &= ~ICRNL;
21:25:22 <aum> newtset.c_cc[VTIME] = 0;
21:25:23 <aum> newtset.c_cc[VMIN]  = 1;
21:25:25 <aum> tcsetattr(STDIN_FILENO, TCSANOW, &newtset);
21:25:45 <aum> that disables line-buffering, and allows single-character fetches
21:25:52 <kc5tja> aum: Does that work in an xterm or rxvt window though?
21:25:58 <aum> sure does :)
21:26:04 <Herkamire> aum: what about keyup events?
21:26:05 <kc5tja> Interesting.  
21:26:13 <aum> Herkamire: don't get greedy :)
21:26:21 <Herkamire> that was my point.
21:26:28 <Herkamire> it's a PITA to get it unbuffered
21:26:28 <aum> google for getch and linux
21:26:33 <kc5tja> Herkamire: You need to set the console into Raw mode for that.
21:26:33 <Herkamire> and I don't think you can get keyups.
21:26:39 <kc5tja> And THAT is something I'm not sure xterm/rxvt supports.
21:26:41 <aum> who really needs keyups?
21:26:52 <aum> i'm happy just to get keystrokes as they happen
21:26:55 <kc5tja> aum: Game programmers, more than anyone else.
21:26:58 <Herkamire> games
21:27:14 <Herkamire> doing anything with the keyboard besides type
21:27:19 <aum> ahh, keyups require some x h4x041ng
21:27:21 <Herkamire> making a humane interface
21:27:34 <kc5tja> Herkamire: Good point.  I forgot about that.
21:28:06 <Herkamire> I want to make herkforth so you can hold down the apple key, type a command and release it to execute the command
21:29:33 <aum> Herkamire: man X
21:29:44 <Herkamire> what good is X going to do me?
21:29:50 <Herkamire> I'm not going to implement the X protocol in herkforth
21:30:09 <aum> there are X apps which let you insert keyboard event handlers into the loop, and intercept whatever key combos you like
21:30:24 <aum> alt key, ctrl key, logo key, key-up, key-down
21:30:40 <kc5tja> Ahh, yes, the whole ICCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCM protocol.  :D
21:30:47 <kc5tja> And I think I forgot a few C's too.  :D
21:30:49 <aum> get the source to xev
21:31:02 <aum> xev demonstrates how to grab low-level key events
21:31:13 <kc5tja> Man, at least Amiga's IDCMP acronym was easy to type.  :D
21:31:25 <slava> i have a new weblog entry describing factor's garbage collector: http://www.jroller.com/page/slava/
21:31:38 <aum> warning though - low level X programming is like DIY root canal therapy with a rusty hand-drill
21:31:41 <Herkamire> I'd rather write a C program that uses SDL, and run the forth image under that
21:32:14 <Herkamire> aum: I'm not touching X
21:32:14 <aum> X libs are written by that school of programmers who say "if it's hard to write, it should be hard to use"
21:32:29 <aum> Herkamire: that might save your sanity :)
21:32:38 <slava> well sdl is a simple framebuffer, X is network graphics -> much more complex.
21:32:52 <Herkamire> I don't give a crap about network graphics
21:32:59 <slava> then don't use X
21:33:02 <slava> use the sdl api
21:33:25 <kc5tja> aum: Actually, X11 was written by the folks who thought that X10 compatibility was paramount.  Then, on top of that, X11R2 all the way up to X11R6, should be backward compatible.
21:33:41 <Herkamire> I am also not going to make herkforth be able to link to libraries.
21:33:42 <kc5tja> Much of the cruft in X11 was simply taken out of the W window manager's API.  
21:34:22 <kc5tja> see, that's where Herkamire and I will differ.  I will link against C libraries.  Someday.  Not now, but someday... :D
21:34:32 <slava> Herkamire, why not?!
21:34:35 <Herkamire> yeah, we have different goals
21:34:39 <Herkamire> I'm making an operating system
21:34:46 <Herkamire> not a handy linux programming environment
21:36:05 <Herkamire> doing the things I want to do under linux would be impossibly complex
21:36:39 <slava> why
21:37:01 <Herkamire> keyboard input is a bastard as I have just described.
21:37:04 <Herkamire> graphics are slow
21:37:19 <slava> how are graphics slow? you can access the fb directly if you want, or use shared memory under x11.
21:37:20 <Herkamire> only way to get decent framerates are with OpenGL with is complex
21:37:32 <Herkamire> yes, I can, and I get about 10fps
21:37:33 <slava> i'd rather use x11 and opengl than write a keyboard driver :)
21:37:36 <Herkamire> I want 70fps
21:38:09 <Herkamire> I promise my keyboard driver will be simpler than trying to interact with opengl and X11
21:38:24 <slava> except you'll have to write 10 keyboard drivers for each architecture.
21:38:34 <Herkamire> no I wont.
21:38:49 <slava> otherwise it will only run on your kit
21:38:55 <Herkamire> almost everybody with a recent mac has a USB Apple Pro Keyboard
21:39:06 <slava> mac is like 3% of all computers
21:39:34 <kc5tja> USB keyboards all follow the same HID class for keyboards in general.  Many of the scan codes are the same as the PC USB keyboards.
21:39:44 <Herkamire> probably more than the hardware compatible with colorforth
21:39:51 <kc5tja> (which borrows from the standard PS/2 keyboard scancode set 0 IIRC).
21:39:56 * aum still fails to understand why apple strangled themselves by hanging on to their hardware monopoly
21:40:01 <slava> ok, but if you forgo linux you'll have to write your own disk driver/graphics driver/etc too.
21:40:13 <kc5tja> aum: Because they can't compete with Microsoft on a software-only basis.
21:40:30 <Herkamire> aum: don't know. perhaps because it's way better... <shrug> that's why I like it anyway.
21:40:33 <aum> kc5tja: they were way ahead of M$ in 1984 when the mac came out
21:40:53 <kc5tja> aum: In 1884, EVERYONE was holding onto their hardware.
21:40:57 <kc5tja> 1984 rather
21:41:18 <Herkamire> apple wants to deliver consistancy and ease of use
21:41:20 <aum> kc5tja: bullshit, 1983 - IBM released the first PC, hardware specs were open, clones were permitted
21:41:26 <Herkamire> it can't do that if everybody has a different crappy bios
21:41:36 <kc5tja> aum: Because IBM did not foresee large sales.
21:41:41 <Herkamire> apple opened up to clones for a while, and it was a disaster.
21:41:45 <aum> kc5tja: I question that
21:41:52 <Herkamire> the clones required special system disks, wouldn't start off the CD with the C key
21:42:00 <kc5tja> aum: It's well known that if IBM had predicted the PC to take off as big as it has, it would have been more tightly controlled (witness the PS/2 line).
21:42:06 <Herkamire> some required their own special firmware upgrades...
21:42:15 <slava> i'm looking forward to using a nice clean architecture, without decades of cruft... the forthbox kestrel :-)
21:42:25 <kc5tja> The PS/2 is arguably vastly superior to even today's PC architectures with PCI bus.  Yet, it's a dead architecture.
21:42:29 <aum> talk about kruft
21:43:15 <kc5tja> (And frankly, the whole PS/2 line came about because of, you guessed it, NOT the Macintosh, but the Amiga!)
21:43:20 <aum> love that P4, a hack of a P3, hack of P2, hack of Pentium, hack of 486, hack of 386, hack of 8086, hack of 8088, hack of 8080, hack of 4040, hack of 4004
21:43:49 <kc5tja> as I recall, the 4040 never existed; only the 4004 was made.
21:43:53 <kc5tja> But otherwise, yes.
21:44:17 <Herkamire> ppc instruction set seems well designed to me
21:44:19 <kc5tja> And the 8088 is a hack of the 8086 (same core; different BIU circuitry).
21:44:36 * aum started hacking in '76, with an 8-bit board with LEDs and toggle switches and 256 bytes RAM
21:44:47 <kc5tja> Herkamire: That's because IBM virtually invented RISC (well, they DID invent RISC; just not the RISC microprocessor -- see the IBM System/800 back in the late 70s).
21:45:29 <kc5tja> IBM S/800 -> IBM PC/RT -> IBM RS/6000 (POWER-I) -> PowerPC
21:45:35 <Herkamire> I wish there was a post-incrementing store... but that's my only complaint so far
21:45:45 <kc5tja> The lineage is readily apparent -- many of the instruction set conventions date to the great mainframes.
21:45:57 <kc5tja> Yeah.
21:46:15 <Herkamire> oh, and the way the first two operands are swapped for "logical" instructions.
21:46:39 <kc5tja> What do you mean by logical instructions?
21:48:20 <Herkamire> and, or, shift etc
21:48:23 <Herkamire> I think
21:48:33 <Herkamire> not in the asm syntax, but the way they pack into the instructions
21:50:25 <Herkamire> iirc most instructions put the destination register first (right after the opcode). "logical" instructions put a source register first, then the destination
21:50:39 <kc5tja> ???
21:50:43 <kc5tja> That's bizarre.
21:50:47 <Herkamire> yes
21:50:55 <kc5tja> The assembler gets it right though, right?
21:51:12 <Herkamire> yeah. the asm syntax always has destination first
21:51:36 <kc5tja> That at least is a good thing.
21:51:57 <Herkamire> eg add is encoded: [ opcode | destination | src1 | src2 | opcode2 | 0 ]
21:53:02 <Herkamire> and is encoded: [ opcode | src1 | destination | src2 | opcode2 | 0 ]
21:53:18 <kc5tja> I wonder what caused them to do that.
21:53:27 <Herkamire> no idea
21:54:53 <Herkamire> I guess I only had to implement two extra words in herkforth to handle it so far, but it just seems silly.
21:55:21 * kc5tja nods
21:55:27 <kc5tja> I'm sure IBM had a good reason though.
21:55:39 <kc5tja> They are simply not known for making maniacle technological decisions.
21:56:00 <kc5tja> (even the IBM PC made perfect sense *at the time* and for the conditions they thought would occur.)
21:58:54 <Herkamire> pegasos looks like a cool system. I don't know all that much about hardware though.
21:59:10 <Herkamire> (seemed to have less cruft to me)
22:00:29 * kc5tja isn't too familiar with the hardware.
22:00:42 <kc5tja> But the software sounds nice.  Basically an OS that is almost a direct descendent of AmigaOS.
22:00:55 <poingie> gar. amigaos
22:01:32 <Herkamire> and it runs linux (gentoo I think) which in turn can run Mac OS if you like that sort of thing
22:02:12 <Herkamire> I haven't started up MOL in months though.
22:02:34 <kc5tja> I heard MOL was amazingly good.
22:02:47 <kc5tja> A bit on the slow side, but still nonetheless very good.
22:02:58 <Herkamire> works OK for me. not great
22:03:09 <Herkamire> but it may be that my mac os installation has issues.
22:03:26 <Herkamire> I would not be suprised if it was mac issues not issues with MOL
22:12:16 --- quit: snowrichard (Nick collision from services.)
22:12:28 --- join: richard_ (richard@adsl-068-209-159-248.sip.shv.bellsouth.net) joined #forth
22:12:33 --- join: snowrichard (richard@adsl-068-209-159-248.sip.shv.bellsouth.net) joined #forth
22:13:07 --- quit: richard_ (Client Quit)
22:13:16 <snowrichard> hello
22:13:28 <kc5tja> Howdy
22:13:44 <snowrichard> how is your emulator coming along?
22:14:20 <kc5tja> The emulator itself is not being worked on.  Rather, I'm concentrating on the basic ROM software that will ship with it.
22:14:29 <kc5tja> The assembler now supports 'defer, 'is, and ''
22:14:30 <snowrichard> ok
22:14:44 <slava> cool
22:14:50 <kc5tja> (so now I can compile code in the various vectors without muddling up the sources)
22:15:16 <kc5tja> I also fixed at-xy to not leave cursor droppings either (and added two new words, "show" and "hide", which shows and hides the text cursor.  Yes, they nest.)
22:15:44 <kc5tja> My next step I think is to work on TYPE and somehow get the assembler to recognize and compile strings.
22:16:07 <kc5tja> Not yet sure how I want to represent strings in memory just yet.  Packed strings are nice, but difficult to create, I think.  I still need to experiment.
22:16:52 <snowrichard> unicode characters might be half a cell wide anyway
22:18:30 * slava is debugging some java code, it ain't fun ;)
22:18:49 <slava> luckily its my own code, not some random library.
22:18:59 <snowrichard> using the sun sdk?
22:19:36 <slava> yup
22:20:01 <snowrichard> I downloaded J, which is a descendant of APL, and it needed java for the gui
22:20:14 <snowrichard> ide gui that is
22:20:25 <slava> why didn't they write it in j?
22:21:45 --- join: foobar (fooboo@dialup-4.152.162.131.Dial1.Atlanta1.Level3.net) joined #forth
22:22:31 <snowrichard> I got out of the hospital wednesday and I am still picking glue from the tape they used from my arm.
22:25:02 <snowrichard> ~=== break 
22:26:06 <kc5tja> Hehe
22:26:29 --- join: lalalim_ (~lalalim@p508ABD87.dip.t-dialin.net) joined #forth
22:26:53 <kc5tja> slava: Not sure.  K's GUI is written in C, but that's because it's an intrinsic part of the language.
22:26:57 <kc5tja> J is a darn cool language though.  :D
22:27:37 <kc5tja> I need to learn more of it, but it has greatly influenced my CPU's design, and the design of my Forth environment I'm working on.
22:28:10 <slava> how so?
22:28:27 <snowrichard> I read a few books about APL a long time ago, and stole some time on a terminal, but didn't stay on it long
22:28:37 <snowrichard> selectric typewriter apl terminal
22:28:50 <kc5tja> well, it made me realize the importance of arrays.  Here's a language that doesn't even tokenize its input source, interprets it naively, yet offers performance comparable with unoptimized C.
22:28:54 <kc5tja> That's damn impressive.
22:29:00 <snowrichard> that dates me some I guess _)
22:29:24 <slava> not pre-tokenizing is a design flaw
22:29:24 <kc5tja> Hehe :D
22:29:38 <kc5tja> slava: No, it's not.
22:29:53 <kc5tja> How do you tokenize +, ., /, *, :, ?, =, [, etc?
22:30:00 <kc5tja> If you've ever seen J source code, you'll understand.
22:30:16 <snowrichard> looks like line noise at first
22:30:21 <foobar> Is K better than J?
22:30:26 <slava> you replace characters with pointers to their code definition et
22:30:27 <slava> etc*
22:30:28 <kc5tja> foobar: Yes, in many ways.
22:30:35 <slava> pre-parse escapes in string
22:30:35 <kc5tja> foobar: However, J is easier to learn, I've found.
22:30:57 <snowrichard> I have my browser tuned to the ANS forth draft
22:31:25 <kc5tja> slava: When interpretation time accounts for less than 0.1% the average run-time of an application, who cares about tokenization?
22:31:45 <slava> kc5tja, in a tight loop you don't want to keep parsing the same text over and over
22:32:15 <kc5tja> slava: In J, you don't do that anyway.  It's an array language -- it natively supports implicit iteration.
22:32:18 <snowrichard> you don't need as many explicit loops in APL or J
22:32:38 <slava> still you might need an explicit loop at some stage and it will be inefficient.
22:33:44 <foobar> Here's a spreadsheet in six lines of K: http://www.kochandreas.com/home/language/cases/SPREADSHEET_K.HTM
22:33:53 <kc5tja> J and K are both used to manipulate databases larger than the combined address space of all the PCs in the state of California.  They don't use explicit iteration.
22:33:55 <snowrichard> my first language was Fortran IV
22:34:34 <slava> foobar, that is unreadable
22:34:52 <kc5tja> It's unreadable because you don't know K.
22:35:00 <kc5tja> Forth is totally intractable to most C programmers.
22:35:02 <snowrichard> the college mainframe had 512 kbytes of core
22:35:07 <slava> `$' is not a good name for a function.
22:35:11 <slava> what is :[;
22:35:39 <kc5tja> slava: Instead of dismissing the langauge on the basis of its readability, I would recommend reading some of the tutorials first.
22:35:58 <kc5tja> I am getting offended at the off-handed dismissal of the language.
22:36:09 <snowrichard> K or J?
22:36:14 <snowrichard> or L?
22:36:17 <snowrichard> :)
22:37:12 --- quit: snowrichard ("Leaving")
22:37:28 <foobar> if one used the language daily, '$' and ':[;' would become second nature, i suppose
22:37:38 * Tomasu is away: night
22:37:48 --- nick: Tomasu -> TomasuDlrrp
22:37:49 <kc5tja> Yes.
22:39:46 --- quit: lalalim (Read error: 110 (Connection timed out))
22:42:55 <foobar> K can be downloaded here: http://www.kx.com/download/download.htm
22:43:51 <kc5tja> I've used K before; I actually kind of prefer J over K -- for some reason, it seems easier to use to me, despite being a more complex language.
22:44:01 <kc5tja> But, my preferred language is still Forth.
22:44:38 <kc5tja> One of my daydreams is to make a library of Forth words that provides much of the same capabilities as J and/or K.
22:45:14 <slava> you'd need gc or refcounting for that
22:45:49 <foobar> J capability in Forth sounds good
22:46:06 <slava> a vector stack?
22:46:16 <kc5tja> J doesn't do GC of any kind (I'm not sure how they accomplish this; perhaps linear memory management), but I know K does reference counting, which it can do because it can guarantee everything is tree structured.
22:46:59 <slava> refcounting is kind of slow though. but simple to implement
22:47:08 <kc5tja> foobar: That is largely why my CPU has three address registers and six memory accessors.  :)
22:47:25 <kc5tja> slava: If I were to implement a vector package, it'd use real GC -- Oberon style.
22:47:38 <slava> what kind of gc does oberon do?
22:48:28 <kc5tja> Mark and sweep, but it uses type descriptors to identify where pointers lie.  But in a vector (unlike a K list), all items are of the same type, so the vector itself can identify what kind of vector it is.
22:48:41 <slava> ok.
22:49:01 * kc5tja was considering mark-n-compact, which would be ideal.  But, it would require a lot of indirection.
22:49:15 <kc5tja> Well, not once you're working inside a vector, but to get TO a vector that is.
22:49:17 <slava> stop and copy is the simplest to implement but there's the 2x overhead.
22:49:31 * kc5tja nods
22:50:08 <kc5tja> It's good for generational collectors though, as I understand, since the younger generations are collected so frequently, that the overhead of other techniques dwarfs the time to copy.
22:50:20 <kc5tja> I guess it depends on the target CPU though.
22:50:23 <slava> yup.
22:50:31 <slava> also copying gc makes allocation very simple.
22:50:36 <slava> no need to maintain a free list.
22:50:53 <kc5tja> A compacting GCer is a single-pool variant of a copy GCer.
22:51:01 <slava> yup.
22:51:07 <kc5tja> That's why I was going with a compacting GCer.
22:51:25 <foobar> stupid question: in ANS Forth, if you use ALLOCATE you must use FREE before you exit the program, right?
22:51:57 <kc5tja> foobar: Depends on the host OS.  Many OSes will automatically free your memory.  Some, like AmigaOS, will not.  It's best if you FREE it though.
22:52:08 <foobar> ok
22:52:54 * kc5tja will be right back.
22:58:51 --- quit: poingie ("Chatzilla 0.9.64b [Mozilla rv:1.7/20040626]")
23:04:30 --- quit: htp123 ("*")
23:05:48 <slava> "After 67-year-old Hurshell Ralls went into surgery for bladder cancer, he came out of surgery missing more than he ever expected. His penis and testicles were gone. "
23:06:50 <foobar> another reason to shun the medical establishment
23:13:49 <kc5tja> back
23:15:49 <kc5tja> Damn, I'm getting tired.
23:15:54 <kc5tja> At least I'm at a good stopping point.
23:16:20 <kc5tja> Although it's not much progress, it is definitely some.
23:24:09 <Herkamire> I little progress every day get's you places
23:25:11 <Herkamire> I think most of the herkforth is made up of small advancements
23:26:34 <slava> here is a quote about java -vs- perl for web stuff
23:26:37 <slava> "For the past 6 years I've been selling an application (the ad server that runs here on Javalobby) written in Perl and over the course of the last two years I've rehashed it in Java. Let me tell you, there is a huge difference: Java without question is faster and more scalable. With the Perl version running under mod_fastcgi, we were lucky if a Dual Xeon 2 GHz with 2 GB of RAM could serve more than 3 million ads per da
23:26:37 <slava> y. The Java version on the same hardware has proven itself by pretty effortlessly serving 9 million ads per day."
23:29:48 <Herkamire> so, run the perl version under mod_perl
23:31:42 <ianp> network latency 
23:31:56 <ianp> java is surprisingly fast enough
23:33:15 <Herkamire> well, that's a new one on me.
23:33:23 <Herkamire> I've never ever seen a java program do anything quickly
23:33:47 <slava> i've never seen a forth program do anything *useful*
23:33:50 * slava ducks
23:34:10 <Herkamire> hehe :)
23:34:12 * foobar searches for his gun
23:34:45 <Herkamire> I've... translated hex <--> decimal...
23:35:37 <foobar> many Forthers defected to Jave: Tom Zimmer, Wil Baden, Bill Muench, etc.
23:35:41 <foobar> java
23:35:50 <Herkamire> foobar: weird
23:37:05 <foobar> a message from Baden was recently forwarded to c.l.f. in which he said Forth was only for computers with "constrained resources"
23:37:05 <Herkamire> someday... I should write a juggling simulator for my OS :)
23:37:51 <Herkamire> whereas java is only for computers with too many resources?
23:38:08 <foobar> i guess
23:39:21 <Herkamire> gosh, these processors are getting so fast. we can cut the speed down 100x by running everything on a virtual machine. oh boy ;)
23:40:06 <Herkamire> it'll be portable, and run equally crappy on the very few systems that support java well
23:40:13 <Herkamire> and usually not at all on other systems
23:40:46 <foobar> i wish the ANS standard had made forth more competitive with java by include a lot more string-handling words
23:41:03 <Herkamire> if java's so freaking portable, then why does it usually not work on my computer?
23:41:34 <Herkamire> string handling can easily be done by including some other file
23:41:47 <Herkamire> it would be nice to have a standard for graphics
23:41:56 <foobar> yes
23:42:11 <Herkamire> I have no problem writing my own string functions.
23:42:40 <foobar> but it's easier to use pre-written standard functions
23:42:41 <Herkamire> but I generally don't want to muck deep enough into other people's forth, and the host system to get graphics working
23:42:58 <foobar> as in C, Awk, Icon, basic, etc.
23:43:44 <foobar> the only free forth i could find for graphics under DOS is Smal32 from russia
23:44:35 <Herkamire> someday someday someday
23:45:08 <Herkamire> I'll have my own forth os
23:45:10 <Herkamire> good drivers
23:45:18 <Herkamire> simple to program for and change
23:46:06 <Herkamire> I really want to play with graphics, but it really grinds me how much linux slows the graphics down
23:47:38 <Herkamire> damn. I'm getting payed to set a linux machine up to replace one, and get the timesheets database back up.
23:47:51 <Herkamire> I got the system, webserver database etc intsalled.
23:47:59 <Herkamire> but I can't find the data for the database.
23:48:07 <Herkamire> once I plunk that it in should be done
23:48:34 <foobar> good luck
23:59:59 --- log: ended forth/04.07.09