#jsmess 2019-03-31,Sun

↑back Search

Time Nickname Message
00:10 🔗 arkiver has quit IRC (Read error: Operation timed out)
00:10 🔗 balrog has quit IRC (Read error: Operation timed out)
00:11 🔗 logchfoo2 starts logging #jsmess at Sun Mar 31 00:11:27 2019
00:11 🔗 logchfoo2 has joined #jsmess
00:16 🔗 arkiver has joined #jsmess
00:23 🔗 db48x has joined #jsmess
00:31 🔗 datajerk has joined #jsmess
01:13 🔗 godane has joined #jsmess
02:03 🔗 balrog has quit IRC (Read error: Operation timed out)
02:07 🔗 balrog has joined #jsmess
03:57 🔗 balrog has quit IRC (Write error: Broken pipe)
03:57 🔗 balrog has joined #jsmess
03:57 🔗 datajerk has quit IRC (Read error: Operation timed out)
04:02 🔗 datajerk has joined #jsmess
04:40 🔗 DFJustin has quit IRC (Remote host closed the connection)
04:57 🔗 DFJustin has joined #jsmess
10:52 🔗 godane has quit IRC (Ping timeout: 255 seconds)
11:06 🔗 godane has joined #jsmess
14:47 🔗 SketchCow Bummer
14:47 🔗 SketchCow db48x: Also sad on wasmer
18:48 🔗 azakai has quit IRC (Read error: Operation timed out)
20:54 🔗 godane has quit IRC (Ping timeout: 252 seconds)
20:55 🔗 godane has joined #jsmess
21:15 🔗 azakai has joined #jsmess
21:57 🔗 SketchCow DFJustin: What gets fixed then - MAME or Emscripten compiling
22:03 🔗 DFJustin mame
22:04 🔗 DFJustin I have a fix that makes tapper work but I'm refining it before putting it in
22:35 🔗 SketchCow OK
22:36 🔗 SketchCow The problem is all Bally/Midways, basically
22:36 🔗 SketchCow So I assume it's all the same driver issues
22:36 🔗 SketchCow Also, I'm still focusing up our spreadsheet of works/doesn't work
22:36 🔗 SketchCow https://docs.google.com/spreadsheets/d/1r_fpzrlUMS4A3IjlbVnA6jNbcSv6rgfz4B-RwCEVDPY/edit?usp=sharing
22:36 🔗 SketchCow I'll be on that after I finish editing my podcasts today
22:37 🔗 DFJustin https://github.com/mamedev/mame/commit/293fb66d7f56897f587721254e47d855241d514d
22:37 🔗 DFJustin should fix mcr.cpp and cchasm.cpp and maybe others
22:41 🔗 db48x https://milek7.pl/openttd-wasm/ has really good sound
22:42 🔗 db48x DFJustin: of course, to get that fix we have to figure out why full-screen is broken in MAME 0.206
22:42 🔗 db48x the inner dimensions of the canvas end up being 0x0
22:46 🔗 SketchCow Agreed on the fullscreen issue, but also glad that we're going to see improvements on the crashes
22:46 🔗 SketchCow db48x: The next question is how to set it up so you can compile a recent MAME and we can test it
22:48 🔗 db48x we're already all set up to compile it
22:49 🔗 db48x ok, I've checked out that commit and started a build
22:49 🔗 db48x it'll only build mcr, so it won't take very long
22:55 🔗 * db48x waits impatiently
22:56 🔗 DFJustin fullscreen worked on the channelf build I did the other day but doesn't on my mcr build
22:57 🔗 DFJustin it looks like SDL is reporting the screen size as 0x0
22:59 🔗 DFJustin hmm no
23:18 🔗 SketchCow db48x: I mean test it so we can run two different mame*
23:18 🔗 SketchCow Like from emularity_engine_candidate
23:19 🔗 db48x so slow:
23:19 🔗 db48x [2019-03-31 22:50:49] Building src/mame/drivers/mcr3.cpp into mamemcr3.js.gz
23:19 🔗 db48x [2019-03-31 23:15:09] Building src/mame/drivers/mcr3.cpp into mamemcr3_wasm.js.gz
23:19 🔗 db48x that's about 10x slower than it used to be :()
23:20 🔗 db48x SketchCow: I can upload a new json file which points to the new js and wasm, and then we can change the root beer tapper item to use the new json file
23:20 🔗 SketchCow Yes
23:20 🔗 SketchCow But I meant a system maybe in the loader.js
23:20 🔗 db48x or I could just upload it as-is, since root beer tapper doesn't work anyway
23:20 🔗 SketchCow So, in the VERY SPECIFIC CASE of rbtapper, just overwrite
23:20 🔗 SketchCow Since yeah, it's fuckedd
23:20 🔗 db48x yea, a general mechanism for testing would be pretty nice
23:21 🔗 SketchCow But I meant in the future, to do that x/year compile, then we do all the calls to make sure things are woring.
23:22 🔗 bai automated browser testing is certainly possible, but it's been a while since I set anything like that up and things have changed since then
23:22 🔗 DFJustin 25 minutes seems slower than what I get, are you rebuilding the whole thing for the wasm version or just the final step
23:23 🔗 SketchCow I think it's as simple as a second object in the loader.
23:23 🔗 SketchCow so that if you say test=1 or whatever, it uses emularity_candidate_v1
23:23 🔗 SketchCow So we compile to that second set, with everything, and then when done, aim the uploads back at engine_v1
23:24 🔗 SketchCow (Since config_v1 and bios_v1 and the rest are not affected by MAME except in a pure fixing sense)
23:24 🔗 db48x DFJustin: wasm just finished. it took 8 minutes, so I guess the first one was long because there were 2k new commits
23:25 🔗 SketchCow "@Jason Scott Is it normal for the games to only use partial of the screen's canvas?
23:25 🔗 bai I'm not sure it even needs to be hosted. with a testing framework, you could set it up to just run a series of tests once the compile is done, on the build system, just cranking through however many thousand items we want tested using headless chrome and/or ff, and give you a nice spreadsheet of red or green for each test
23:25 🔗 SketchCow Going to full screen works fine though"
23:25 🔗 SketchCow bai: I'd agree if I didn't think we needed to do it by hand
23:25 🔗 SketchCow Because the problems are myriad
23:26 🔗 bai yeah, some problems are easier to detect than others
23:26 🔗 SketchCow Did it boot, did it error, did it do wrong aspect ratio, did sound stop
23:26 🔗 SketchCow You could do an automated OH IT DEFINITELY BLEW THE FUCK UP
23:26 🔗 bai this would catch severe regressions like "throws an exception while loading"
23:26 🔗 SketchCow EVER'Y TING BLOWING UP MON
23:26 🔗 bai you can write tests for more detailed stuff but yeah, it gets trickier
23:27 🔗 SketchCow Here is what I know
23:27 🔗 SketchCow Almost all the errors in the sheet, right now, have been there for weeks and weeks.
23:27 🔗 SketchCow So, I'm inclined to do the human version
23:27 🔗 SketchCow And then, sure, if you want to take some of that VR Helmet Braniac and burn it on improved automated testing, great
23:28 🔗 db48x has quit IRC (Read error: Connection reset by peer)
23:28 🔗 db48x has joined #jsmess
irclogger-viewer