00:00:00 --- log: started retro/11.02.19
05:01:55 --- join: qbject (~qb@229.sub-174-254-1.myvzw.com) joined #retro
05:25:54 --- mode: ChanServ set +o crc_
05:25:59 --- nick: crc_ -> crc
06:41:44 --- quit: qbject ()
07:21:19 * crc has some updates to the javacript vm up at http://rx-core.org/proto/
07:40:49 <crc> this allows the canvas to be toggled on/off, and also the Words.txt can be viewed
07:44:35 <docl> cool!
07:45:33 <docl> I'm not happy with the speed at which the text comes up when I type words; it takes 5 seconds or more to read out the whole thing.
07:54:43 <crc> I'll work on that
07:55:08 <docl> thanks :)
08:36:00 <crc> docl: try it now
08:37:54 <docl> much faster, more like 1-2 seconds
08:39:03 <crc> and once more :)
08:40:44 <docl> still taking about a second. also there is an effect where part of the screen goes black for a split second.
08:44:40 <crc> which browser, os, and how fast is your cpu?
08:45:17 <docl> firefox, windows 7, slow netbook (lemme look it up)
08:46:13 <docl> the blank screen seems to happen when words hits the end of the text box and needs to scroll down
08:46:31 <docl> (just a portion of the screen goes blank, black to be precise)
08:47:29 <crc> ok. the vm is probably injecting html into the output div faster than firefox can keep up with it. I'll see if I can find a way to fix this
08:48:34 <docl> Intel Atom CPU N450 @ 1.66GHz
08:48:41 <docl> 1.00GB
08:48:50 <docl> 32-bit operating system
08:49:04 <crc> thanks
08:49:21 <docl> it would be a freaking supercomputer if it weren't for modern software bloat! :P
08:53:44 <crc> the speed of output is tricky to get right. I can make it really fast, but that eats into battery life on mobile devices, and forces you to type slowly (or keyboard events get lost)
08:56:32 <crc> it's still a lot better than the original few tries, which took ~15 minutes to execute words ;)
08:58:12 <docl> :)
09:00:48 <docl> I'm irritated that I can't increase the screen's virtual size on my netbook and zoom around to whatever corner I want, or zoom out like in a print preview... this could be so much better.
09:19:22 <crc> bbl; closing up at work now
12:02:29 --- quit: scj (Read error: Connection reset by peer)
12:27:10 --- join: scj (scj@mycrosoft.us) joined #retro
12:35:43 --- quit: scj (Disconnected by services)
12:38:26 --- join: scj (scj@mycrosoft.us) joined #retro
14:59:22 --- join: roarde (~roarde@pdpc/supporter/active/sixforty) joined #retro
15:50:30 --- quit: roarde (Quit: disgusted)
16:20:51 --- join: roarde (~roarde@pdpc/supporter/active/sixforty) joined #retro
16:34:35 <roarde> I give up. How do I open a file for read?
16:40:47 <crc> with files'
16:40:47 <crc> "filename" :R open
16:51:22 <roarde> I wouldn't accept that "1" was the handle. Finally decided to open a second file.
17:08:32 <crc> you can have 12 files open on the VM's that support file I/O; valid handles will vary by VM
17:08:43 <roarde> k
17:09:27 <roarde> that's a port limit and not a limit of file's implementation, right?
17:11:12 <crc> the limit is imposed by the VM implementations. Nothing wrong with increasing the limits :)
17:12:52 <roarde> in general, what kind of shape is the dynamic vm in?
17:13:10 <crc> it's partially broken at the moment
17:13:17 <roarde> ok
17:13:29 <crc> I'll try to fix it tonight
17:41:03 <roarde> is there a retro.rx vm?
17:42:15 <crc> no, though an ANS implementation is coming
17:44:08 <crc> implementing the vm in retro seems slightly pointless to me; it'd be unable to load full images, and likely run too slowly to be practical
17:45:12 <roarde> I'm looking at compiling one using posix libs.
17:49:12 <roarde> The only advantage would be quicker understanding for those more familiar with Retro than with anything else.
17:49:12 <roarde> For that to be of help, system library docs would have to be referenced in included texts.
17:49:12 <roarde> So yeah, the more one looks at it, the less it appeals.
17:50:10 <crc> fwiw, updated benchmark results (including the unreleased ANS implementations from cfa) are at https://spreadsheets.google.com/pub?key=0ArQ9snhGsa7HdHpubXdRTFZEbDNoR2RuY0kwc21tbFE&output=html
17:55:09 <crc> gonga (the go implementation from yiyus) will be added as soon as I can finish building go on the server
18:02:26 <crc> done, and updated
18:11:55 <crc> roarde: the dynamic vm appears to work properly now
18:12:12 <roarde> ok, thanks
18:19:20 <roarde> wow. feeding -O3 to gcc makes an extreme amount of difference on this box. Rebuilding image takes about 2/3 the time.
18:52:16 <crc> good to know
18:53:10 <roarde> a lot of quirks with platform; one example is that MMX is disabled, tho present. Many similar problems.
18:55:51 <crc> su
19:20:21 <crc> I added clang, tcc, and pcc to the benchmark chart to show relative performance on my test box across four different C compilers
19:23:04 <roarde> pcc's not as disappointing as I feared. I'm thinking of setting it up as the system compiler, as I'm a bit of a license geek.
19:28:04 <crc> roarde: updated again with "pcc -O vm/retro.c -o retro" to use optimizations
19:29:43 <crc> not quite as good as gcc -O3, but looks promising
19:30:42 <roarde> actually, pretty impressive. Thanks for doing that.
19:30:49 <crc> not a probl
19:30:55 <crc> *problem
20:39:35 --- join: pietro10 (~pietro10@ool-4a591559.dyn.optonline.net) joined #retro
20:39:54 --- part: pietro10 left #retro
20:46:00 --- quit: roarde (Quit: Leaving.)
23:59:59 --- log: ended retro/11.02.19