Nightly crashes when trying to use a patched version of Howler.js to test Web Audio API support.

Do you have a link to the crash report? You should be able to get one from about:crashes. Thanks!

I posted some repro code here: https://github.com/canuckistani/howler.js/tree/bug_845960_createGain In particular I added this change to get the library to try to use createGain instead of chrome's createGainNode: https://github.com/canuckistani/howler.js/commit/4dfd9583da240806db09e02da25df605452638d8#howler.js

Here's the crash report link: https://crash-stats.mozilla.com/report/index/bp-d82a7784-3e92-404a-b8f4-be5822130227

Thanks, my first guess is that we're decoding an MP3 which fails and then we hit something bad in the failure code... I'll take a look.

This bug was almost certainly caused because I had forgotten to put the audio files in place. Now that I have done so, I get different errors from Howler.

(In reply to Jeff Griffiths (:canuckistani) from comment #3) > https://crash-stats.mozilla.com/report/index/bp-d82a7784-3e92-404a-b8f4- > be5822130227 It's a 2-week old Nightly.

D'oh, you're exactly right. I've updated, here is a new crash report: https://crash-stats.mozilla.com/report/index/bp-3e4cfca8-99e5-4421-b03b-9f2f72130227

(In reply to comment #5) > This bug was almost certainly caused because I had forgotten to put the audio > files in place. Now that I have done so, I get different errors from Howler. Oh, actually I realize I never tested this with an empty input buffer, I don't think. :-)

Should be pretty easy to fix then :) If you go in for this sort of thing, the user story might be something like: "As a developer, if I try to play an audio file using AudioContext but the file is not available, the api should throw an exception that I can catch in a try block."

(In reply to comment #9) > Should be pretty easy to fix then :) If you go in for this sort of thing, the > user story might be something like: > > "As a developer, if I try to play an audio file using AudioContext but the file > is not available, the api should throw an exception that I can catch in a try > block." Oh yeah, in fact our code does try to call the failure callback function provided by the user to decodeAudioData, it just crashes in the process of doing that! :-)

speex_resampler_process_*'s input buffer is not affected by its channel_index argument, hence the bug. I'll need to adjust the tests for the fix. Patch coming up.

(In reply to :Ehsan Akhgari from comment #11) > speex_resampler_process_*'s input buffer is not affected by its > channel_index argument, hence the bug. I'll need to adjust the tests for > the fix. Patch coming up. Wrong bug. :(

So I just looked at this. I cloned the branch in comment 2 and when I open test/index.html in that file and press play I cannot reproduce the bug...

(See this URL for example: http://people.mozilla.com/~eakhgari/howler.js/test/) Jeff, do you still see the crash?

(In reply to :Ehsan Akhgari from comment #14) > (See this URL for example: > http://people.mozilla.com/~eakhgari/howler.js/test/) > > Jeff, do you still see the crash? With your example, it seems to be successfully loading sound.ogg. try deleting that, or changing the code to: var sound = new Howl({ urls: ['foo.ogg'], autoplay: false, loop: true, volume: 0.7, onend: function() { console.log('Finished!'); } });

OK great, now I can reproduce the crash!

Very easy to fix, I was just missing a null check.

Comment on attachment 721046 [details] [diff] [review] Patch (v1) r=me

Mass moving Web Audio bugs to the Web Audio component. Filter on duckityduck.