00:00:00 --- log: started forth/02.12.06
00:27:51 --- quit: Serg_Penguin ()
02:06:00 --- join: Klaw (chuck@ip68-4-243-214.oc.oc.cox.net) joined #forth
06:08:50 --- join: Herkamire (~jason@wsip68-15-54-54.ri.ri.cox.net) joined #forth
06:12:35 <Robert> Hi Herkamire.
06:29:33 <Herkamire> Hi Robert :)
06:29:39 <Herkamire> the start of a new day :)
06:30:39 <Robert> Really?
06:30:46 <Robert> Looks more like a grey afternoon to me.
06:32:30 <Herkamire> hey, whatever works for you ;)
06:34:47 --- join: tathi (~josh@wsip68-15-54-54.ri.ri.cox.net) joined #forth
09:39:17 <Herkamire> dylan actually works on ppc :)
09:41:31 <proteusguy> lucky guy
09:53:38 --- quit: ian (forward.freenode.net irc.freenode.net)
09:53:40 --- join: ian (ian@inpuj.net) joined #forth
10:33:40 --- join: Speuler (~l@mnch-d9ba45c4.pool.mediaWays.net) joined #forth
10:34:27 <Speuler> g'day
10:48:35 <Robert> Hej Speuler.
11:21:00 <Speuler> hi robert
11:21:07 <Speuler> be greeted
11:21:13 --- join: ASau (~asau@158.250.48.197) joined #forth
11:21:28 <Speuler> #openbios has a new topic line
11:21:46 <Speuler> now it says:
11:21:49 <ASau> Good evening!
11:21:58 <Speuler> "Forth spoken here"
11:23:57 <Robert> :)
11:25:02 <Speuler> t'is an attempt to implement an open version of open firmware
11:25:22 <Speuler> open firmware has been written in forth, btw
11:25:34 <Speuler> so openbios will be forth too
11:25:47 <Speuler> their interpreter is called "paflof"
11:25:48 <ASau> I see 3 of 7 people on #openbios that are on #forth.
11:26:03 <Speuler> well, two just dropped in
11:26:09 <Speuler> namely herk and you
11:26:10 <ASau> Do they explain such an interest in forth
11:26:40 <Speuler> by need, yes
11:27:02 <Speuler> as you must be able to program open firmware in forth ...
11:27:54 <ASau> It is an interesting thing, 4/5 are not afraid of using Forth, it seems.
11:28:36 <Speuler> their interpreter, as far as it is working now, is on:
11:29:09 <ASau> One more Forth in C?
11:29:45 <Speuler> http://forthfreak.net/openbios/paflof-dump-20021016.tar.gz
11:29:52 <Speuler> ASau: C++ even
11:29:56 <Speuler> must be portable
11:30:07 <ASau> Brrr...
11:30:33 <Speuler> as openbios is supposed to support non-intel too
11:30:48 <Speuler> specifically alpha
11:31:11 <ASau> Don't they have asm for alpha?
11:31:30 <ASau> Or can't they write it?
11:31:54 <Speuler> would be difficult to maintain different code bases while it is still worked on
11:32:35 <Speuler> the interpreter is a forth subset
11:32:45 * Robert is now on a Minix box... Too bad I haven't seen any Minix forths.
11:33:04 <Speuler> Robert: you have gcc
11:33:16 <Speuler> but you can also ask asau to write you an asm version
11:33:26 <ASau> One reason why I liked FIG Forth model is it doesn't claim compatibility, but shows the way to easy ports.
11:33:53 <Speuler> as easy as saying "make" ?
11:34:01 <Robert> Isn't the gcc for Minix port rather old=
11:34:08 <Robert> Like...2.7.something.
11:34:28 <Speuler> well, it worked then. it should do so now
11:35:01 <ASau> If you can say "make", somebody had written the code before you.
11:35:20 <Speuler> right. so you don't need to write it again
11:35:22 <ASau> If you have to port, then you have to write this code.
11:35:41 <ASau> One thing I hate in FSF sources:
11:36:07 <Speuler> the idea of paflof is to pull as much as possible out, and reimp it in forth
11:36:26 <Speuler> so paflof is just a lib of some primitives
11:36:38 <Speuler> at that point, asm porting should be much less work too
11:36:55 <ASau> I want sources for _my_ platform only, why shoul I get all that machine dependencies for MIPS/Alpha/PPC/etc.?!
11:36:57 <Speuler> that's the work going on now
11:37:24 <Speuler> ASau: because you would have to write it first. the alpha sources wouldn't run on your box
11:38:03 <ASau> It shold be done in separate add-ons/patches.
11:38:17 <Speuler> sure. you're free to do so
11:39:13 <Speuler> "oh shit this bug doesn't show in the alpha version"
11:39:29 <Speuler> or
11:39:44 <Speuler> "i've enhanced the ppc version a lot, please port it"
11:39:53 <ASau> If I take gforth as example, I have about a half Meg wasted by machine dependent code for all CPU-s developpers know.
11:40:24 <Speuler> ASau: you mean, asm files ?
11:40:47 <ASau> Asm and CPU dependent C.
11:41:00 <Speuler> ASau: how would you avoid those by writing in assembly language ?
11:41:11 <ASau> If I take gcc, I can't imagine what shall I get.
11:41:37 <Speuler> porting gforth takes about 24 hours
11:41:43 <ASau> Have you seen A. Horsts ciforth?
11:41:59 <Speuler> i know albert, but not his forth
11:42:40 <ASau> You have many text macropreprocessors.
11:43:13 <ASau> If you want to process Forth to asm, you don'n need much power.
11:44:13 <Speuler> ASau: why don't you complain about the zillions of files for each supported architecture, like you do with gforth ?
11:45:01 <ASau> Well, that's a matter of principle.
11:45:54 <ASau> If one don't claim easy portability and volume&speed quality,
11:46:22 <ASau> I don't expect them to be the fact.
11:46:57 <ASau> But Forth community, I mean, esp. some well-known advocates,
11:47:11 <ASau> claim these.
11:48:16 <ASau> So I should wonder, why did they make such desicions on design.
11:48:35 <Speuler> because somebody just wrote it
11:48:47 <Speuler> and there was no need to port it
11:49:03 <Speuler> because it could just be recompiled for intel
11:49:25 <Speuler> now it's going to be stripped
11:49:33 <Speuler> and c++ replaced against forth
11:49:43 <Speuler> rather than asm
11:50:02 <Speuler> or would you object to using forth instead of asm ?
11:50:29 <ASau> I think it is possible to use forth instead of asm.
11:50:39 <Speuler> to an extend, yes
11:50:55 <Speuler> somewhere there must be cpu-executable code
11:51:22 <Speuler> and, when generating that code, knowledge about the cpu must be built into the compiler
11:52:16 <ASau> If I want to get a code for forth interpreter, I can use classical threading
11:52:28 <Speuler> ASau: primitives too ?
11:52:41 <Speuler> how would you write  "drop" ?
11:52:44 <ASau> and rewrite less than 32 primitives in code directly.
11:53:04 <Speuler> ASau: for paflof, 0 primitives have been rewritten
11:53:27 <ASau> Just as I wrote directly in code for KP580BM2 CPU.
11:53:46 <Speuler> it took me 5 seconds to "port" paflof
11:55:01 <ASau> If you are designer/programmer of paflof, you possibly can.
11:55:17 <ASau> But not a 3rd man.
11:55:17 <Speuler> nope. i just downloaded it and ran "make"
11:55:32 <ASau> Oh, you are lucky.
11:55:42 <ASau> It looks like this.
11:56:12 <ASau> I once saw how a man tryed to compile something from sources.
11:56:29 <ASau> IIRC, that was MPlayer for Linux.
11:56:37 <Speuler> i did so yesterday too
11:56:59 <Speuler> (new version)
11:57:44 <ASau> Buildability of MPlayer under linux didn't save it to be got rid of.
11:58:04 <Speuler> rephrase last sentence please
11:58:23 <ASau> So, I have to conside such portability as a fairy tale.
11:58:42 <Speuler> because you've seen somebody fail ?
11:58:58 <Speuler> i once saw somebody fail his driver license.
11:59:08 <Speuler> so driver licenses are fairy tales
11:59:35 <Herkamire> Speuler: hehe
12:00:34 <ASau> I see too many fails in compiling FSF code in cases, that are claimed by developpers not to bring any problems.
12:01:00 <ASau> That is a kind of new ms-windows.
12:01:13 <ASau> Just "make" and play. :)
12:01:58 <Speuler> ASau: no matter what your arguments are, you can't discount the fact that the openbios forth interpreter runs on several cpus without the need to reimplement it or parts of it in assembler
12:02:07 <ASau> That's why I consider succeded recompilation from sources as a matter of luck.
12:03:03 <ASau> Sometimes I'm lucky too. :)))
12:03:26 <Speuler> the mplayer problem could have been solved by reimplemting it in assembler, i guess
12:03:59 <ASau> No. I shan't say this.
12:04:14 <Speuler> you stated the problem, now state a solution
12:04:21 <ASau> I say, that C is not a good choice to write portable programs.
12:04:30 <Speuler> assembler is ?
12:04:38 <ASau> No.
12:04:43 <Speuler> basic ?
12:05:28 <ASau> My first math. teacher used to repeat a proverb:
12:05:54 <ASau> "Every vegetable has it own time."
12:06:56 <Speuler> ASau: what vegetable is good for porting code ?
12:07:19 <ASau> If there exist such P.L. as Cobol and Fortran, that doesn't mean they are good.
12:07:39 <Speuler> i didn't claim so
12:08:05 <Speuler> but what IS good for portable code ?
12:08:11 <ASau> I add, even if there is much code written.
12:08:46 <ASau> I'd looked at Ada, Oberon, Modula for imperative programming.
12:09:04 <Speuler> i'd say, that language is good for coding for which you got a compiler on the target platform
12:09:29 <Speuler> s/coding/porting/
12:09:41 <Speuler> or interpreter
12:09:49 <Speuler> or translator
12:09:49 <ASau> Of course. No doubt.
12:10:31 <ASau> But I see today much is done to use C/C++ where it is appropriate and write it is not.
12:11:10 <ASau> Just like Perl code to dump a word from fixed displacement in binary file.
12:14:57 <ASau> There is another point.
12:15:17 <ASau> I think such large sources are useless too.
12:16:05 <ASau> And the case of obtaining sources is like obtaining binary code.
12:16:27 <ASau> With all its magic, "make"-and-play.
12:16:51 <ASau> If it doesn't make, you have much to reverse engeneer.
12:17:19 <ASau> About 11 MB gzipped in case of gcc.
12:18:12 <Speuler> we got a common opinion there
12:18:27 <Speuler> i too believe it is important to write compact code
12:18:33 <Speuler> not just the binary output
12:18:40 <Speuler> but the sources too
12:19:11 <Speuler> i'll leave you at this
12:19:21 <Speuler> because i'll jump out for a while
12:19:31 <Speuler> there'e "tollwood" around the corner
12:19:52 <ASau> Our FIDOnet writers used to cry in such moments:
12:19:54 <Speuler> which is an alternative multicultural bazaar
12:19:59 <ASau> "Consensus!!!"
12:24:21 <ASau> "Time, forth!" (G.Sviridov.)
12:24:24 <ASau> Bye!
12:24:29 --- part: ASau left #forth
12:32:01 --- quit: TreyB ()
12:32:42 --- join: lament (~lament@h24-78-145-92.vc.shawcable.net) joined #forth
12:48:22 --- join: TreyB (~trey@cpe-66-87-192-27.tx.sprintbbd.net) joined #forth
13:27:43 --- join: I440r (~I440r@sdn-ap-029tnnashP0253.dialsprint.net) joined #forth
13:27:56 --- join: Klaw` (chuck@ip68-4-243-214.oc.oc.cox.net) joined #forth
13:28:43 --- part: Klaw` left #forth
13:39:33 --- join: wossname (wossname@HSE-QuebecCity-ppp81313.qc.sympatico.ca) joined #forth
14:18:18 --- quit: Herkamire ("Off for the weekend")
14:22:12 --- quit: tathi ("leaving")
14:38:30 --- quit: I440r (Read error: 110 (Connection timed out))
14:43:35 <Speuler> re
15:04:19 --- join: miraage (~miraage@h24-78-143-48.vc.shawcable.net) joined #forth
15:04:32 --- part: miraage left #forth
15:11:23 --- join: jamc (~user@as3-6-8.asp.s.bonet.se) joined #forth
15:12:48 <Robert> Hej du :)
15:14:17 <jamc> hej du med
16:23:04 --- quit: ChanServ (Shutting Down)
16:23:17 --- join: ChanServ (ChanServ@services.) joined #forth
16:23:17 --- mode: forward.freenode.net set +o ChanServ
16:23:17 --- mode: ChanServ set +l 83
16:49:43 --- quit: Fractal (Read error: 110 (Connection timed out))
17:07:55 --- join: tathi (~josh@ip68-9-68-215.ri.ri.cox.net) joined #forth
17:11:38 <XeF4> tathi: how did you deal with the disappointment of discovering dcba is unimplemented on ppc750?
17:11:52 * XeF4 needs some help in coping
17:14:37 --- quit: jamc ("SANTA CLAUS comes down a FIRE ESCAPE wearing bright blue LEG WARMERS..  He scrubs the POPE with a mild soap or detergent for )
17:21:36 <tathi> XeF4: dcba?
17:21:48 * tathi goes to find his PPC reference manual...
17:22:03 <XeF4> data cache block allocate
17:22:56 <tathi> hmm...I've never really done anything performance intensive enough to need to get into cache management much
17:23:11 <tathi> what's that do, tell it to load a line from main memory into the cache?
17:23:29 <XeF4> tells it to assign an EA to a cache line without loading from main memory
17:23:42 <XeF4> nice for building temporary lookup tables
17:23:47 <tathi> ah, right
17:24:02 <XeF4> er rather, to assign a cache line to an EA
17:25:35 <tathi> how much of a performance boost does that give you?
17:26:47 <tathi> hmm, I'm not much help, am I? :)
17:27:55 <XeF4> not 100% certain, but for allocating scratch area for raytracing, it looks like it could save a few memory stalls anyway
17:28:06 <XeF4> haven't profiled it yet
17:28:37 <tathi> ok, I can see that
17:31:27 <tathi> hmm, if your data page is caching-allowed, dcbz might help you...
17:34:43 <tathi> unless that's unimplemented also, I don't have 750 docs on hand
17:37:45 <XeF4> it is, but it is 6 cycles, which is worse than loading from L2
17:37:53 <XeF4> or.. maybe not quite.
17:39:56 <XeF4> and it serializes execution
18:02:54 --- join: bwb (~bwb@ip68-4-124-131.oc.oc.cox.net) joined #forth
18:30:36 --- join: Fractal (ksf@h24-77-171-228.ok.shawcable.net) joined #forth
18:32:41 --- join: tcn (tcn@tc2-login35.megatrondata.com) joined #forth
18:40:11 --- quit: Speuler (Read error: 110 (Connection timed out))
18:41:27 --- join: Speuler (~l@mnch-d9ba4882.pool.mediaWays.net) joined #forth
19:31:09 --- quit: tathi ("leaving")
19:33:52 --- quit: Speuler (Read error: 110 (Connection timed out))
19:36:51 --- quit: wossname ("yes, evening all..")
20:38:13 --- quit: tcn ()
23:28:33 --- quit: XeF4 (Remote closed the connection)
23:59:59 --- log: ended forth/02.12.06