00:00:00 --- log: started forth/04.01.03
00:44:25 --- join: arke (~chris@wbar8.lax1-4-11-099-104.dsl-verizon.net) joined #forth
01:45:49 --- join: schihei (~schihei@pD9548583.dip.t-dialin.net) joined #forth
01:55:14 * arke is away: sleep
03:03:04 --- join: qFox (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
03:06:07 --- quit: qFox (Read error: 104 (Connection reset by peer))
03:06:11 --- join: qF0x (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
07:11:20 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth
07:11:49 --- quit: tathi (Client Quit)
07:18:01 --- part: aktnot_ left #forth
08:24:32 --- join: schihei_ (~schihei@pD9E5CC11.dip.t-dialin.net) joined #forth
08:32:44 --- quit: schihei (Read error: 60 (Operation timed out))
08:52:14 --- join: kc5tja (~kc5tja@66-91-231-74.san.rr.com) joined #forth
08:52:15 --- mode: ChanServ set +o kc5tja
09:24:08 --- join: savage- (~savage@twistedmatrix.com) joined #forth
09:24:30 --- part: savage- left #forth
09:48:56 --- quit: kc5tja ("THX QSO ES 73 DE KC5TJA/6 CL ES QRT AR SK")
11:13:07 --- quit: schihei_ (Client Quit)
11:51:04 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth
11:51:40 <Robert> Hi Herkamire :)
11:56:50 * arke is back (gone 10:01:38)
11:57:57 <Herkamire> hi Robert :)
12:06:59 --- quit: onetom (Read error: 110 (Connection timed out))
12:12:37 * warpzero is back (gone 14:23:52)
12:55:50 * arke is away: errands
13:01:57 <slava> hi Herkamire
13:04:41 <Herkamire> hi slava :)
13:05:16 <warpzero> Walk like a champION talk like a champION
13:05:18 <slava> Herkamire, i'm working on an http server written in my scripting language. it does server side scripting now
13:05:48 <Herkamire> cool :)
13:06:08 <Herkamire> slava: the LSD language?
13:06:22 <slava> Herkamire, go to http://slava.kicks-ass.org:8888/dictionary.lhtml
13:06:42 <slava> Herkamire, this is a dynamically-geneated list of all defined words in the current interpreter; click on one to see the decompiled definition
13:07:03 <slava> Herkamire, it doesn't do URL decoding yet, so it won't work with words whose names contian special characters...
13:07:05 <Herkamire> lol! nice URL :)
13:13:47 <Herkamire> slava: nice :)
13:14:12 <Herkamire> slava: how hard would it be to make each word in the definitions a link?
13:14:49 <slava> pretty easy
13:15:03 <slava> first i need to do a url decoder, as well as converting < and > to &entities;
13:15:17 <Herkamire> cool
13:15:21 <slava> i put the latest httpd source in index.html
13:15:23 <Herkamire> that's great you wrote an httpd
13:15:53 <Herkamire> I've thought a bit about what I want an httpd to do
13:15:59 <Herkamire> (that apache doesn't)
13:16:37 <slava> this is crazy -- once i add bindings for java.nio (async io api) and also a compiler, this http server should be nice and fast
13:16:54 <slava> right now its main problem is that it only handles 1 connection at once
13:17:44 <Herkamire> yeah, I wrote a couple dummy http servers, and never dealt with concurrent connections
13:17:52 <Herkamire> except through tcpserver
13:22:18 <Herkamire> compiler... you mean compile java to native machine code?
13:25:32 <slava> no, compile lsd to java vm bytecode on the fly
13:25:43 <slava> so there is still a dynamic interpreter, but it runs faster
13:25:58 <Herkamire> ic
13:31:15 <Herkamire> does in run slowly now?
14:10:28 --- join: TheBlueWizard (TheBlueWiz@pc84dn1d.ppp.fcc.net) joined #forth
14:10:28 --- mode: ChanServ set +o TheBlueWizard
14:10:34 <TheBlueWizard> hiya all
14:12:28 <Herkamire> hi TheBlueWizard :)
14:12:39 <TheBlueWizard> hiya Herkamire
14:14:31 <mur> good night
14:14:56 <TheBlueWizard> g'nite, mur
15:03:01 * arke is back (gone 02:07:11)
15:04:56 <Herkamire> arke: don't sleep much?
15:05:09 <arke> !?
15:05:17 <arke> I was just running some errands, thats all
15:05:22 <arke> and missed half of the figuk meeting :(
15:05:28 <Herkamire> crap I gotta look at nicks more closely
15:06:50 <arke> :)
15:22:47 <I440r> the fig uk meeting ?
15:22:49 <arke> irc.stealth.net / #figuk
15:22:52 <arke> come on forthers :)
15:22:55 <arke> yep yep yep
15:22:56 <arke> right now
15:22:57 <I440r> i know where its at.
15:22:58 <arke> :)
15:22:59 <I440r> is it on now ?
15:23:05 <arke> yup yup yup 
16:04:01 * arke is away: afk
16:24:18 * arke is back (gone 00:20:17)
16:52:47 <arke> latest in #figuk: 
16:52:59 <Robert> No need to flood.
16:53:09 <arke> we were discussing how x86 needs to be revamped.
16:53:15 <arke> Robert: I wasn't gonna flood ^____^
16:53:20 <Robert> Lies.
16:53:41 <Robert> Go find out how much  sub esi,byte 4 / mov [esi],eax  costs compared to  push eax
16:53:47 <Robert> In clocks, that is.
16:53:53 <Robert> On a modern x86.
16:54:12 <arke> nnkay, I'll try to find several CPUs
16:54:14 <arke> including AMD
16:54:39 <Robert> Guess it won't be THAT much of a difference. Just don't look at HelpPCs cycle counts for the 8086.
16:55:28 <arke> lol
16:55:39 <arke> it might actually be more than you think, since its pipelined.
16:56:02 <arke> however, it might not, depends on how b0rked the x86 pipeline is with reads before writes
16:59:27 <Robert> I was refering to the difference between AMD and Intel CPUs.
17:01:02 <Robert> Did you find out?
17:01:51 * warpzero is away: Let's get a party goin', let's get a part goin', let's get a party goin', party hard, party hard: irc.wsyntax.com
17:02:59 <arke> still looking around.
17:03:06 <arke> its harder to find than one might think.
17:03:06 <Robert> OK :)
17:03:07 <arke> gimme a hand
17:03:08 <Robert> Hehe.
17:03:15 <Robert> Nah, I was going to do some homework.
17:05:16 <arke> ugh
17:05:17 <arke> lolk
17:05:27 <arke> intel optimization manual does not have the clocks
17:05:47 <Robert> Oh, well...
17:06:24 <Robert> I guess you still save a bit compared to doing  lodsd / jmp eax  every instruction.
17:06:37 <Robert> And "nest" being a single call instruction.
17:06:51 <Robert> Even though working with the stack becomes harder.
17:06:53 <arke> ugh
17:06:58 <arke> yes, STC is definetely faster.
17:07:17 <Robert> If you could do those register optimizations, it would certainly go a lot faster!
17:08:01 <arke> Heh, I can do the inliner, but that one I probably couldn't do...
17:08:08 <arke> unless I found a good example of it somewhere
17:08:19 <Robert> Right, I guess it's rather complex.
17:08:38 <Robert> Allthough experimenting with it might be fun.
17:09:19 <Robert> Remember that saving several register to the stack will go faster.
17:09:29 <Robert> sub esi,byte 4*n
17:09:40 <Robert> mov [esi+0*4],reg1
17:09:42 <Robert> mov [esi+1*4],reg2
17:09:46 <Robert> mov [esi+n*4],regn
17:10:03 <Robert> Or maybe it's (n-1)*4.. you get the idea.
17:10:43 <arke> dunno about that, since repeated write to the same register (even if indirect) will still cause pipe muck
17:11:18 <Robert> Eh, you only write to esi once.
17:11:22 <Robert> That's the whole point.
17:11:34 <Robert> After that you only access memory and the other registers.
17:13:58 <arke> oh yeah, true.
17:14:05 <arke> but it still mucks once
17:14:10 <arke> because of the read-after-write.
17:14:25 <arke> so better optimization is to stick sub esi, byte 4 somewhere randomly inbetween
17:14:53 <Robert> Or afterwards.
17:14:59 <Robert> Then it'll be write-after-read.
17:15:31 <arke> thats better, actually.
17:15:31 * Robert is stupid.
17:15:40 <arke> ^___^
17:15:46 <arke> actually, that would be much better.
17:23:49 <Robert> Hmm..tomorrow I will experiment with that thing.
17:24:04 <arke> sweet.
17:24:05 <arke> but
17:24:08 <arke> I'm not done yet.
17:24:09 <arke> :(
17:25:52 --- quit: qF0x (Read error: 104 (Connection reset by peer))
17:26:15 <Robert> I mean, the register thingie.
17:26:23 <Robert> Doesn't seem very hard after all!
17:28:32 <arke> how so?
17:28:45 <arke> it seems way friggin hard, especially in a stack based language...
17:29:33 <Robert> Well...
17:29:45 <Robert> I'll use some pseudo-C code.
17:30:01 <arke> register->stack and stack->register conversions are hell.
17:30:05 <arke> Robert: !?
17:30:13 <Robert> string reg[4] = { ecx, edx, ecx, edi }
17:30:23 <Robert> int index = 0;
17:30:37 <Robert> Now, for example, compiling the literal "123" would be:
17:30:49 <Robert> mov reg[index++],123
17:30:55 <Robert> I.e.  mov ecx,123
17:31:14 <Robert> Let's do it again, but with the literal 666
17:31:18 <Robert> mov reg[index++],666
17:31:23 <Robert> I.e.  mov edx,666
17:31:26 <Robert> Addition would be:
17:31:44 <Robert> add reg[i-2],reg[i-1]
17:31:48 <Robert> i = i - 1
17:31:52 <Robert> I.e.  add ecx,edx
17:31:55 <arke> Hrm.
17:31:57 <Robert> And edx is freed.
17:32:02 <Robert> Now here's the beatiful part:
17:32:04 <arke> that requires multi-step compilation though :((
17:32:14 <Robert> drop can be optimized to:
17:32:16 <Robert> i = i - 1
17:32:23 <Robert> I.e.  no machine code!
17:32:25 <Robert> swap becomes:
17:32:40 <arke> xchg :)
17:32:40 <Robert> exchange(reg[i-2],reg[i-1])
17:32:49 <Robert> I.e. no machine code either!
17:32:49 <arke> so xchg ^__^
17:32:52 <Robert> No.
17:32:55 <arke> er
17:32:55 <arke> oh
17:32:56 <arke> yeah
17:32:59 <Robert> That's done compile-time.
17:33:09 <Robert> dup and over are each one instruction.
17:33:12 <arke> you're a genius
17:33:24 <Robert> I.e.  reg[i] = reg[i-1];  i++
17:33:32 <Robert> Er
17:33:38 <Robert> mov reg[i],reg[i-1]
17:33:39 <Robert> i++
17:33:40 <Robert> That is.
17:33:58 <Robert> And whenever you call a function, you do:
17:34:09 <Robert> while(i > 0) {
17:34:37 <Robert>  mov [esi-..something],reg[i]
17:34:39 <Robert>  i--
17:34:39 <arke> thatss sweet!!!
17:34:40 <Robert> }
17:34:45 <arke> keep going, i'll brb
17:40:35 <Robert> A for/next loop would be:
17:40:39 <Robert> for:
17:41:01 <Robert> Allocate a register (index j) from the other end of the register list.
17:41:14 <Robert> mov reg[j],reg[i-1]
17:41:14 <Robert> i--
17:41:19 <Robert> Do stuff...
17:41:21 <Robert> next:
17:41:34 <Robert> dec reg[j]
17:41:42 <Robert> jnz address_to_for
17:41:49 <Robert> deallocate reg[j]
17:43:45 <Robert> Just keep the number of stack items used constant in the loop ;)
17:43:52 <Robert> Same thing about begin/until
17:44:50 <Robert> Hrm. One problem though - the number of stack items used in a certain function has to be known.
17:45:54 <Robert> Er.. actually just the number of return values and parameters, now when I think of it.
17:52:01 <arke> well, actually, they all need to be knows :((
17:52:15 <Robert> Hm?
17:52:18 <Robert> What do you mean?
17:53:21 <arke> <Robert> Hrm. One problem though - the number of stack items used in a certain function has to be known.
17:53:29 <arke> needs to know how many it uses in the beginning as well.
17:54:11 <Robert> No.
17:54:18 <Robert> Or how do you mean?
17:54:34 <Robert> Everything the calling word uses is pushed.
17:54:57 <Robert> All you need to do is to cache the parameters.
17:54:57 <arke> exactly.
17:55:00 <arke> Oh.
17:55:02 <arke> I know.
17:55:06 <Robert> i=0
17:55:08 <arke> if it needs it, insert it at the end.
17:55:09 <arke> lol
17:55:11 <arke> ^__^
17:55:18 <Robert> mov reg[i++],[esi]
17:55:20 <Robert> mov reg[i++],[esi+4]
17:55:21 <Robert> etc.
17:55:27 <Robert> add esi,byte x
17:55:49 <arke> what happens if it runs out of registers though? (especially on an intel :( )
17:55:57 <Robert> You have 6 registers.
17:56:03 <Robert> That's a lot.
17:56:27 <Robert> Even if you use a for loop, you still can have 5 stack items.
17:56:37 <Robert> That's enough to do something like... 1 2 3 4 5 + + + +
17:56:45 <arke> eax, ebx, ecx, edx, edi, ebp
17:56:50 <arke> 6...
17:56:52 <arke> I guess you're right.
17:56:54 <Robert> Yep.
17:57:08 <arke> the TOS always being cached first.
17:57:18 <Robert> Right.
17:57:24 <arke> I should implement o: and o; in isforth, just to try it.
17:57:33 <arke> er...
17:57:39 <arke> nevermind, won't work in isforth :((
17:57:42 <Robert> No.
17:57:51 <arke> damn DTC.
17:57:53 <Robert> Write a new Forth.
17:58:07 <Robert> Or a simple Forth program outputing optimized asm code.
17:58:11 <Robert> Just to test the idea.
17:58:33 <arke> or just finish F and use that.
17:58:37 <arke> ^______________________________^
17:59:04 <arke> I have lots to do before wednesday.
17:59:15 <arke> I have to completely write a term paper, finish F, do the register thing.
17:59:16 <arke> ^__6
17:59:55 <Robert> :)
18:01:01 <arke> when do you usually wake up?
18:01:10 <arke> (your time)
18:01:16 <Robert> 2pm ;)
18:01:21 <Robert> But I'll set the clock
18:01:29 <arke> (because you fall asleep at 4am :))) )
18:01:50 <arke> aah, ok.
18:01:59 <Robert> Might have something to do with it..
18:02:06 <arke> 2pm your time is 5am my time.
18:02:19 <arke> well, we can talk tomorrow then ^___^
18:02:26 <arke> I'm gonna let my bro use the computer
18:02:27 <arke> later
18:02:29 --- quit: arke ("Leaving")
18:02:29 <Robert> Probably.
18:17:19 --- join: Sonarman (~matt@adsl-64-171-255-248.dsl.snfc21.pacbell.net) joined #forth
18:31:03 <TheBlueWizard> gotta go...bye all
18:31:03 --- quit: Sonarman (Read error: 54 (Connection reset by peer))
18:31:08 --- part: TheBlueWizard left #forth
18:37:28 --- join: imaginator (~gps@166.70.196.201) joined #forth
19:32:44 --- quit: imaginator (".")
19:42:07 --- join: kc5tja (~kc5tja@66-91-231-74.san.rr.com) joined #forth
19:42:07 --- mode: ChanServ set +o kc5tja
20:46:41 --- quit: Herkamire (sterling.freenode.net irc.freenode.net)
20:46:41 --- quit: I440r (sterling.freenode.net irc.freenode.net)
20:46:41 --- quit: cmeme (sterling.freenode.net irc.freenode.net)
20:46:41 --- quit: skylan (sterling.freenode.net irc.freenode.net)
20:46:47 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth
20:46:47 --- join: I440r (~mark4@12-160.lctv-a5.cablelynx.com) joined #forth
20:46:47 --- join: cmeme (~cmeme@216.184.11.30.unused.swcp.com) joined #forth
20:46:47 --- join: skylan (~sjh@vickesh01-4902.tbaytel.net) joined #forth
20:47:51 --- quit: Herkamire (sterling.freenode.net irc.freenode.net)
20:47:51 --- quit: skylan (sterling.freenode.net irc.freenode.net)
20:47:51 --- quit: cmeme (sterling.freenode.net irc.freenode.net)
20:47:51 --- quit: I440r (sterling.freenode.net irc.freenode.net)
20:48:13 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth
20:48:13 --- join: I440r (~mark4@12-160.lctv-a5.cablelynx.com) joined #forth
20:48:13 --- join: cmeme (~cmeme@216.184.11.30.unused.swcp.com) joined #forth
20:48:13 --- join: skylan (~sjh@vickesh01-4902.tbaytel.net) joined #forth
20:50:06 --- quit: Herkamire (sterling.freenode.net irc.freenode.net)
20:50:06 --- quit: skylan (sterling.freenode.net irc.freenode.net)
20:50:06 --- quit: cmeme (sterling.freenode.net irc.freenode.net)
20:50:06 --- quit: I440r (sterling.freenode.net irc.freenode.net)
20:50:28 --- join: Herkamire (~jason@h000094d30ba2.ne.client2.attbi.com) joined #forth
20:50:28 --- join: I440r (~mark4@12-160.lctv-a5.cablelynx.com) joined #forth
20:50:28 --- join: cmeme (~cmeme@216.184.11.30.unused.swcp.com) joined #forth
20:50:28 --- join: skylan (~sjh@vickesh01-4902.tbaytel.net) joined #forth
21:45:52 --- quit: I440r ("Leaving")
21:48:06 --- join: Sonarman (~matt@adsl-66-124-255-54.dsl.snfc21.pacbell.net) joined #forth
21:49:42 --- join: I440r (~mark4@12-160.lctv-a5.cablelynx.com) joined #forth
21:59:17 --- quit: fridge ("Client exiting")
23:01:55 --- quit: kc5tja ("THX QSO ES 73 DE KC5TJA/6 CL ES QRT AR SK")
23:05:49 --- join: Sonarman_ (~matt@adsl-66-124-254-7.dsl.snfc21.pacbell.net) joined #forth
23:08:37 --- quit: Sonarman_ (Client Quit)
23:16:21 --- quit: Herkamire ("bedtime")
23:17:30 --- quit: Sonarman (Read error: 110 (Connection timed out))
23:17:59 --- join: onetom (~tom@cab.bio.u-szeged.hu) joined #forth
23:29:15 --- join: arke (~chris@wbar8.lax1-4-11-099-104.dsl-verizon.net) joined #forth
23:29:55 <arke> hiya
23:59:59 --- log: ended forth/04.01.03