00:00:00 --- log: started retro/12.10.28
00:28:27 <tangentstorm> woohoo! 10 passing tests, showing that retro.py, retro.c, and retro.pas all work the same way for 8 opcodes. :)
00:30:45 <tangentstorm> pushed to lp:~tangentstorm/retro-language/docwork  under test/ngaro
00:30:55 <tangentstorm> i think i'm gonna call it a night. later all.
00:30:58 --- quit: tangentstorm (Quit: leaving)
00:43:17 --- quit: karswell (*.net *.split)
00:43:22 --- quit: foucist (*.net *.split)
00:43:29 --- quit: oPless (*.net *.split)
00:43:42 --- quit: ivan`` (*.net *.split)
00:44:37 --- quit: yiyus (*.net *.split)
00:44:37 --- quit: saper (*.net *.split)
00:53:32 --- quit: docl (*.net *.split)
01:31:34 --- join: docl (~docl@unaffiliated/docl) joined #retro
01:41:01 --- join: foucist (~foucist@ps14150.dreamhost.com) joined #retro
01:41:01 --- join: ivan`` (~ivan@unaffiliated/ivan/x-000001) joined #retro
01:41:01 --- join: karswell (~coat@93-97-29-243.zone5.bethere.co.uk) joined #retro
01:41:01 --- join: saper (saper@wikipedia/saper) joined #retro
01:41:01 --- join: oPless (~oPless@lart.doosh.net) joined #retro
01:41:01 --- join: yiyus (1242712427@je.je.je) joined #retro
01:59:47 --- join: obobo (~chatzilla@dyn-dsl-ly-98-124-43-42.nexicom.net) joined #retro
01:59:47 --- join: timg (~timg@pool-71-180-13-77.tampfl.fios.verizon.net) joined #retro
02:08:45 --- quit: obobo (*.net *.split)
02:08:54 --- quit: timg (*.net *.split)
02:11:09 --- join: obobo (~chatzilla@dyn-dsl-ly-98-124-43-42.nexicom.net) joined #retro
02:11:09 --- join: timg (~timg@pool-71-180-13-77.tampfl.fios.verizon.net) joined #retro
02:34:45 --- quit: karswell (*.net *.split)
02:34:48 --- quit: foucist (*.net *.split)
02:34:54 --- quit: oPless (*.net *.split)
02:35:11 --- quit: ivan`` (*.net *.split)
02:35:38 --- quit: yiyus (*.net *.split)
02:35:48 --- quit: saper (*.net *.split)
02:38:53 --- join: foucist (~foucist@ps14150.dreamhost.com) joined #retro
02:38:53 --- join: ivan`` (~ivan@unaffiliated/ivan/x-000001) joined #retro
02:38:53 --- join: karswell (~coat@93-97-29-243.zone5.bethere.co.uk) joined #retro
02:38:53 --- join: saper (saper@wikipedia/saper) joined #retro
02:38:53 --- join: oPless (~oPless@lart.doosh.net) joined #retro
02:38:53 --- join: yiyus (1242712427@je.je.je) joined #retro
02:39:41 --- quit: SimonRC (Remote host closed the connection)
02:41:25 --- join: SimonRC (~sc@fof.durge.org) joined #retro
02:45:41 --- quit: docl (*.net *.split)
02:56:26 --- join: docl (~docl@unaffiliated/docl) joined #retro
06:23:02 --- join: ncv (~quassel@89.35.216.197) joined #retro
06:23:02 --- quit: ncv (Changing host)
06:23:02 --- join: ncv (~quassel@unaffiliated/neceve) joined #retro
09:59:07 --- join: kumul (~kumul@173.215.130.73) joined #retro
12:37:08 --- quit: timg (Ping timeout: 245 seconds)
12:46:16 --- join: timg (~timg@pool-71-180-13-77.tampfl.fios.verizon.net) joined #retro
12:49:13 --- join: tangentstorm (~michal@108-218-151-22.lightspeed.rcsntx.sbcglobal.net) joined #retro
13:18:44 --- join: Z_Mass (~zachary@bas1-cornwall24-1242467096.dsl.bell.ca) joined #retro
14:08:18 --- join: Mat2 (~claude@91-65-144-133-dynip.superkabel.de) joined #retro
14:08:23 <Mat2> good afternoon
14:10:35 <kumul> afternoon Mat2 :)
14:10:44 <Mat2> hi kumul !
14:10:53 <kumul> i guess this means you wont go to sleep in the middle of a very interesting conversation
14:11:50 <Mat2> sorry, about what ?
14:15:51 <Mat2> i've spend some frustrations with the SDL library
14:19:55 <Mat2> crc: The bitmap fonts in retro 10, was there your own creation ?
14:23:59 --- quit: ncv (Ping timeout: 260 seconds)
14:27:31 <tangentstorm> lol kumul :)
14:27:51 <tangentstorm> what's happening with sdl, mat2?
14:30:47 <Mat2> hi tangentstorm
14:31:19 <Mat2> only some little API changes in the new 2 version
14:31:50 <tangentstorm> I've been working on that ngaro test suite. So far, it tests retro.py, retro.c, and retro.pas... And I've got 18 tests passing on all 3.
14:32:23 <Mat2> and I guess there work identical
14:32:34 <tangentstorm> yep, or the tests would fail!
14:33:14 <tangentstorm> i'm pretty sure there's a problem in retro.py somewhere. it was crashing on me when i was messing with the help text
14:33:28 <tangentstorm> ( but the c version worked with the same code )
14:33:55 <tangentstorm> lots of bugs in my pascal version, but i'm whipping it into shape :)
14:34:16 <Mat2> that can be an implicit side effect of the phyton vm
14:34:51 <Mat2> it's got to read about your progress
14:35:01 <tangentstorm> Mat2: I've already got some character drawing code in javascript and coffeescript
14:35:26 <tangentstorm> you mean python's own virtual machine?
14:35:30 <Mat2> yes
14:36:05 <Mat2> I work on something like a soft processor for text and later graphic generation
14:36:06 <tangentstorm> could be. i don't remember what i did to cause the problem or i'd debug it. 
14:37:31 <Mat2> this would allow a lot more control above simple bitmap based functions
14:38:09 <tangentstorm> Like what?
14:38:09 <Mat2> like resolution changes, blit and screen split effects, palette changes and so on
14:38:37 <Mat2> vsync control
14:38:52 <tangentstorm> interesting
14:39:22 <tangentstorm> I'm still planning to use Aggpas. I'm pretty sure it has truetype in there.
14:39:48 <Mat2> it has
14:40:03 <tangentstorm> it does all its work on arrays, so it should work with any display technology.
14:42:24 <Mat2> I plan to port my vm to at least two boards, the arduinomite and my new raspberri pi
14:42:40 <tangentstorm> Mat2: are you still using the same basic protocol for I/O as ngaro?
14:43:02 <Mat2> no, i'm using a syscall like interface
14:43:23 <tangentstorm> crc mentioned wanting to revamp the i/o protocol
14:43:49 <tangentstorm> where would i look in your code to see it?
14:43:57 <Mat2> the vm should run on bare metal, so I can not use antigrain or cairo
14:44:24 <Mat2> under sources (the interpreter-.. files)
14:47:25 <tangentstorm> i don't understand what i'm looking at :)
14:48:21 <Mat2> look at the mSYS macro, it is at current a stub but I upload the complete interface this week
14:49:10 <Mat2> each parameter execute as independent thread if finished
14:49:12 <tangentstorm> so it would be like   SYS 0   = halt    SYS 1 = clear screen 
14:49:19 <Mat2> yes
14:49:42 <Mat2> SYS 2 = init text mode
14:49:49 <Mat2> SYS 3 = init graphic mode
14:50:05 <Mat2> SYS 4 = init semigraphic mode
14:50:08 <tangentstorm> you said a while back there were 16 opcodes but this looks like a lot more... ?
14:50:53 <Mat2> this is one opcode with an immediate parameter
14:51:31 <Mat2> like LI (load literal)
14:52:23 <tangentstorm> i mean https://www.assembla.com/code/avm/git/nodes/master/sources/interpreter-dtt32.h#ln242
14:52:44 <tangentstorm> is that every combination of 2 ?
14:53:00 <Mat2> yes
14:53:08 <tangentstorm> oh :)
14:53:38 <Mat2> the interpreter fetches instruction bundles, not single instructions
14:54:19 <tangentstorm> so you don't have to do a lot of shifting at runtime. makes sense.
14:54:27 <Mat2> the code format is defined in a way up to 16 instructions can be freely combined for interpretation
14:54:47 <Mat2> my current version bundles 2 instructions
14:57:10 <Mat2> whereby an opcode is a packed bundle of either 8 instruction bundles (64 bit cell size) or 4 (32 bit)
14:59:16 <tangentstorm> neat
14:59:16 <Mat2> so the interpreter only need 1 load operation for 8 or 4 bundles which are cached in a cpu register
15:00:07 <Mat2> it is an ISA for efficient interpretation
15:00:28 <Mat2> and this means minimisation of the interpretation overhead
15:01:14 <Mat2> it really works a bit like a MISC cpu
15:03:00 <Mat2> the whole syscall interface will work asynchron and parallel (good for modern multicore cpus's)
15:09:35 <tangentstorm> i think this is neat but don't know what to do with it yet :)
15:11:52 <Mat2> I plan this as kind of soft core for implementing my own operation environment 
15:12:54 <tangentstorm> i mean i don't know how to fit it in with what i'm doing yet. but i'm interested :)
15:14:20 <tangentstorm> so you may find this interesting
15:14:34 <Mat2> look, if you want to embed an platform independent forth like vm, that is one way to go in an performance effective way
15:15:11 <tangentstorm> yeah, that's why i like it :)
15:16:28 <Mat2> who knows, possibly I found a way one time to reprogram the microcode of these damn Intel cpu's and make it a cheap hardware solution :)
15:16:33 <tangentstorm> haha
15:17:10 <tangentstorm> Mat2: i'm kind of thinking about using your vm with google's NaCl to provide a super fast vm in the browser that isn't javascript
15:18:29 <Mat2> good idea but what is NaCl ? An JIT compiler ?
15:18:33 <tangentstorm> meanwhile: these tests i'm running... they average about 0.114s on my machine for retro.pas ... and 0.876s for retro.c
15:18:36 <tangentstorm> Native Client
15:18:46 <Mat2> ah ok
15:18:49 <tangentstorm> it lets your run intel machine code in google chrome
15:19:02 <tangentstorm> http://code.google.com/p/nativeclient/
15:19:50 <tangentstorm> i don't know why the pascal version is faster. i'm kind of surprised. i expected it to be much slower, looking at the code.
15:20:00 <tangentstorm> maybe just startup time
15:20:07 <Mat2> thanks for the link, that is a good opportunity
15:20:16 <Mat2> do you use freepascal ?
15:20:25 <tangentstorm> yes
15:21:01 <Mat2> feeepascal compiles case constructs very efficient
15:21:04 <tangentstorm> retro.pas uses an array of pointers to procedures for opcode dispatch
15:22:08 <tangentstorm> whereas retro.c is using a case statement
15:22:30 <tangentstorm> i really thought the overhead of the procedure calls would slow things down
15:23:33 <tangentstorm> it may just be startup time though, because it's invoking a new copy of the executable on each test, and there are only a handful of instructions in each image
15:23:35 <Mat2> in C each function is called per double indirection and the function body's for stack frames must be compiled without some obscure optimisations
15:24:26 <tangentstorm> i've also got range checking and stack checking and stuff turned on in pascal
15:25:06 <Mat2> yes but pascal compiler can distinguish procedures from functions
15:25:47 <Mat2> because of the pascal syntax (that was the reason for it)
15:26:46 <Mat2> you can test this, let the compiler generate an assembler listing and look for the stack frames
15:27:52 <tangentstorm> good idea. i'll give it a try after i finish testing all these opcodes.
15:28:56 <Mat2> these NaCl feature is a nice way for web based programming
15:29:19 --- quit: timg (Read error: Connection reset by peer)
15:29:54 <Mat2> hmm, I need a good looking bitmap font
15:33:36 <tangentstorm> https://twitter.com/i/#!/tangentstorm/media/slideshow?url=pic.twitter.com%2FXLVRz0TX
15:34:02 <tangentstorm> that one came from escapes.js http://256.io/escapes.js/
15:36:17 <tangentstorm>  0 5 *   1 5 *   5 5 *     retro.pas says:   -5 -4 0    nice! :)
15:36:23 <tangentstorm> new math
15:36:32 <Mat2> thanks
15:37:23 <Mat2> uh, looks like your code does multiply signed values
15:37:24 <tangentstorm> i bet i copied and pasted the subtraction op :)
15:38:23 <Mat2> sorry add
15:38:33 <tangentstorm> procedure vm.oMUL ; begin data.push( -data.pop + data.pop ) end;
15:38:54 <tangentstorm> yay copy and paste :)
15:39:37 <Mat2> ... brings back the fun of obscure errors :)
15:41:03 <tangentstorm> i must have looked at that 20 times and not seen it 
15:42:18 <tangentstorm> woohoo! and now it actually runs test/base.rx
15:42:27 <Mat2> good
15:43:02 <tangentstorm> oh, but core.rx fails. guess i still have work to do
15:43:30 <Mat2> I think its only a matter of one or two days
15:43:50 <Mat2> can you please give me a link ?
15:44:14 <tangentstorm> to pascal stuff?
15:44:16 <tangentstorm> oh sorry
15:44:41 <tangentstorm> https://github.com/sabren/b4/
15:45:19 <tangentstorm> it's kind of a mess...  retro.pas , ngaro.pas , and ng.*.pas have the vm
15:48:50 <tangentstorm> https://github.com/sabren/b4/blob/master/vm/ngterm.js  has the character generator i made for ngaro.js 
15:49:07 <tangentstorm> somewhere i have a coffeescript version that uses a typed array and should  be a little faster
15:51:19 <Mat2> can I use the font data ?
15:52:36 --- quit: Z_Mass (Quit: Leaving)
15:55:04 <Mat2> oh, that is the VGA rom font I see
15:56:08 <tangentstorm> i think you're right
15:57:51 <Mat2> where have you found it ?
15:59:47 <tangentstorm> i took it from the code for that escapes.js ( an ansi art terminal for javascript )
16:00:00 <tangentstorm> but i think you're right. it's probably originally came from the standard vga font
16:00:39 <tangentstorm> the ngterm.js link i posted at 17:48 has the data at the bottom
16:02:15 <Mat2> I have found the font of the atari800 series
16:02:20 <Mat2> quite nice
16:02:44 <Mat2> but sadly only 7x8
16:07:54 <tangentstorm> i have one i made in high school... don't have a screenshot though. it's just binary.
16:08:33 <tangentstorm> plan is to make a font editor in retroforth as one of my lessons
16:09:09 <tangentstorm> oh! terminus
16:09:45 <Mat2> your font editor will have a happy user: me !
16:11:24 <tangentstorm> http://terminus-font.sourceforge.net/shots.html
16:11:33 <tangentstorm> that ships with debian. it's pretty nice
16:15:59 <Mat2> thanks, I know these fonts from NetBSD but does not know there bundled with debian
16:31:19 <Mat2> time for some sleep
16:31:21 <Mat2> ciao
16:31:28 --- quit: Mat2 (Quit: Verlassend)
17:39:35 <tangentstorm> yay! all 31 opcodes, plus the implicit "gosub" are now tested externally in ngarotest.py, and all 3 implementations with the --dump hook pass! :)
19:23:03 --- join: Z_Mass (~zachary@bas1-cornwall24-1242467096.dsl.bell.ca) joined #retro
20:21:26 --- quit: Z_Mass (Quit: Leaving)
22:45:13 --- quit: kumul (Quit: WeeChat 0.3.9)
23:59:59 --- log: ended retro/12.10.28