#jsmess 2018-09-22,Sat

↑back Search

Time Nickname Message
00:32 🔗 SketchCow sets mode: +oooo azakai db48x godane Sgeo
00:32 🔗 SketchCow sets mode: +o Vito`
00:33 🔗 SketchCow Screenshotter is going, going going
02:57 🔗 Sgeo COBOL 64 requires a SYS 2051 to start
03:16 🔗 davidar has joined #jsmess
03:21 🔗 Ctrl-S has joined #jsmess
03:25 🔗 godane has quit IRC (Ping timeout: 252 seconds)
04:27 🔗 godane has joined #jsmess
16:59 🔗 azakai has quit IRC (Quit: Ex-Chat)
19:00 🔗 SketchCow Sgeo: OK!
19:00 🔗 SketchCow So, I've been enjoying the stuff coming in, we're at 35,200 items or so. It's gonna be great
19:00 🔗 SketchCow Time for QA
19:00 🔗 SketchCow Do you think you can figure out why F12 dies?
19:05 🔗 Sgeo I know exactly why F12 dies. I'm working on it, and have a build that fixes the top level, but some of the submenus are still broken. Also I have to lower the optimization level until I can apply the fix to Emscripten for the bug I encountered, although I don't notice any slowdown at lower optimization level
19:06 🔗 Sgeo Check out F12 here: https://sgeo.github.io/experimental/vice32/x64.html
19:06 🔗 Sgeo Stuff like enabling the status bar from F12 works, stuff like trying to enter a filename does not
19:08 🔗 Sgeo Mostly it should just be a tedious bit of effort to enable each individual not working thing. I have to add the functions to a whitelist to get interpreted. I am able to avoid needing the main loop added to interpreter, unlike DOSBox, at the cost of a potential timing flaw I think
19:38 🔗 bai how are none of our projects on this list? https://github.com/mbasso/awesome-wasm
19:58 🔗 Sgeo Also, check out the Scale2x option in Video settings |> Render Filter
20:03 🔗 Sgeo bai, any particular reason -O1 would be slower to actually build than -O2?
20:06 🔗 bai not really sure, all I know about that comes from https://kripken.github.io/emscripten-site/docs/tools_reference/emcc.html#emcc-o1 and https://kripken.github.io/emscripten-site/docs/optimizing/Optimizing-Code.html
20:07 🔗 Sgeo Ugh I added all these calls to the whitelist but it's still acting like it's in an infinite loop
20:08 🔗 Sgeo bai, I don't add emscripten_sleep to the whitelist, do I?
20:09 🔗 bai uhhh...I can't recall. I think it should be smart enough to do that automatically...
20:20 🔗 Sgeo emscripten_sleep is inherently async, calling a resumption function via setTimeout
21:12 🔗 bai yeah, makes sense
21:18 🔗 Sgeo https://hastebin.com/ducocumeki.rb
21:18 🔗 Sgeo grah
21:19 🔗 Sgeo There's nothing in this stack trace helping me see why it isn't working
21:19 🔗 Sgeo _string_RsDevice1_callback IS in the whitelist
21:19 🔗 Sgeo Note it's a timeout rather than complaint about infinite loop
21:27 🔗 Sgeo ...just tried a build with 100% Emterpretify. It's unusably slow, but also the menu option I was trying STILL doesn't work
21:32 🔗 * Sgeo has a hypothesis
22:17 🔗 Sgeo Hypothesis confirmed: Emterpretify_Async still doesn't like infinite loops that don't contain a function like emscripten_sleep() in them. I had implemented my own SDL_WaitEvent but needed that sleep in there to tell emterpreter to break it. bai, is that documented and I just missed it?
22:27 🔗 Sgeo I am able to type stuff in. Error messages still kill everything. And it looks like it's going to be a royal pain to get all cartridgesto be menu activatable
23:15 🔗 Sgeo Reenabling -O2 with kripken's fix (presumably to be added in a future Emscripten build). I only had -O3 earlier in a pointless attmept to fix sound, shouldn't really be needed
23:47 🔗 Sgeo SketchCow, what are the most important F12 features for us to support?

irclogger-viewer