[00:50] *** kripken_ has quit IRC (Read error: Operation timed out)
[01:07] <SketchCow> DFJustin: db48x could use some help, the build is breaking with MAME again
[01:14] <DFJustin> I'm all ears
[02:17] *** kripken_ has joined #jsmess
[06:26] *** db48x-aeg has joined #jsmess
[06:27] *** db48x-ere has joined #jsmess
[07:00] <DFJustin> built a working pacman with the mame0217 tag and emscripten 1.38.48 (commit c56422590bb911603529052b7bd78e6ca3c9fa27)
[07:04] <DFJustin> looks like there's a major emscripten update so I'll give that a whirl
[07:05] *** db48x-aeg has quit IRC (Ping timeout: 255 seconds)
[07:17] <db48x-ere> I think running wasm2js in the debugger was a mistake
[07:17] <db48x-ere> it seems to be causing a lot of swapping, and the machine is now unresponsive
[07:33] *** db48x-aeg has joined #jsmess
[08:22] <DFJustin> oh goody it throws an exception with upstream 1.39.5 (commit 369013943283939412fb2807bb0d2ded8ebd5a9e)
[08:31] <db48x-ere> which one did you get?
[08:37] <DFJustin> mamepacman.js:16779 exception thrown: 18637712
[08:37] <DFJustin> callMain @ mamepacman.js:16779
[08:37] <DFJustin> not very useful
[08:38] <DFJustin> they probably changed the magic incantation to get debug info
[08:40] <db48x-aeg> -s ASSERTIONS=1, I'm pretty sure
[08:43] <DFJustin> nope I did the usual SYMBOLS=1 ARCHOPTS="-s ASSERTIONS=2 -s SAFE_HEAP=1"
[08:43] <DFJustin> it made a bigger file but no more info is forthcoming
[08:45] <db48x-aeg> hmm
[09:29] <db48x-aeg> oh, I just noticed
[09:29] <db48x-aeg> warning: undefined symbol: emscripten_glBeginQuery
[09:29] <db48x-aeg> that's new
[09:58] <db48x-ere> well, I got a stack trace that time, but it was just for abort(out of memory)
[09:59] <db48x-ere> unfortunately it happened before the usual messages from MAME, so it's not the same error
[10:38] *** db48x-aeg has quit IRC (Read error: Operation timed out)
[15:13] <SketchCow> Calling alon.
[16:50] *** db48x has joined #jsmess
[17:14] *** db48x has quit IRC (Read error: Operation timed out)
[19:15] <SketchCow> Hey, kripken_ 
[19:15] <SketchCow> 02:00 < DFJustin> built a working pacman with the mame0217 tag and emscripten 1.38.48 (commit c56422590bb911603529052b7bd78e6ca3c9fa27)                  
[19:15] <SketchCow> 02:04 < DFJustin> looks like there's a major emscripten update so I'll give that a whirl          
[19:16] <SketchCow> 03:22 < DFJustin> oh goody it throws an exception with upstream 1.39.5 (commit 369013943283939412fb2807bb0d2ded8ebd5a9e)         
[19:16] <SketchCow> 03:37 < DFJustin> mamepacman.js:16779 exception thrown: 18637712               
[19:16] <SketchCow> 03:37 < DFJustin> callMain @ mamepacman.js:16779       
[19:16] <SketchCow> 04:29 < db48x-aeg> oh, I just noticed             
[19:16] <SketchCow> 04:29 < db48x-aeg> warning: undefined symbol: emscripten_glBeginQuery                   
[19:16] <SketchCow> Anyway, sounds like the new changes made things k-razy
[19:16] <SketchCow> Help accepted
[19:16] <SketchCow> 03:43 < DFJustin> nope I did the usual SYMBOLS=1 ARCHOPTS="-s ASSERTIONS=2 -s SAFE_HEAP=1"                         
[19:54] *** kripken_ has quit IRC (Read error: Operation timed out)
[21:28] <DFJustin> the point where the exception occurs is where there is an expected exception as part of the startup process that's supposed to be whitelisted
[21:32] <DFJustin> (judging by the -verbose output at least)
[21:51] <DFJustin> looks like I just need to pass the flags earlier in the compile process
[21:52] *** kripken_ has joined #jsmess
[21:59] <kripken_> hi SketchCow
[22:00] *** balrog has joined #jsmess
[22:05] <kripken_> I don't follow the webgl side of things that closely myself, but looks like glBeginQuery is under WebGL2, so you need to set MAX_WEBGL_VERSION=2. We refactored those flags recently. But the old flag USE_WEBGL2 should also work, so it's odd if it doesn't.
[22:06] <DFJustin> kripken_: I think the issue I'm getting is just a matter of needing to set EXCEPTION_CATCHING_WHITELIST during the codegen phase
[22:06] <DFJustin> after switching from fastcomp to upstream
[22:06] <kripken_> DFJustin: Oh, yes, that's definitely a change with the new backend. Yeah, it does codegen during the compile stage now (like a regular C compiler), so that flags need to be set there too for exceptions etc.
[22:07] <DFJustin> there are a bunch of new undefined gl messages but there were some of those before too that never seemed to matter
[22:21] <kripken_> i saw an issue filed on a wasm2js segfault with MAME earlier today, maybe that was from someone here? I'll take a look later when I'm at my main machine
[22:25] <SketchCow> Yes, that was db48x
[23:34] *** balrog has quit IRC (Bye)
[23:55] *** balrog has joined #jsmess