Some Rust demuxer issues may not be correctly reported, especially cases where the output is different from the C++ demuxer's output. These should be reported so that they make the DecoderDoctor drop-down appear and users may report them as site issues. This should help our developers learn about issues, more easily reproduce them thanks to the extra information, including the resource URL.

Try with bug 1343437 and bug 1343442: https://treeherder.mozilla.org/#/jobs?repo=try&revision=ef348891b4d5602e14604d88f689ad890336faea

Please don't start reviewing yet, something landed in m-c that will require me to do an extensive rebase! Please wait for the next push.

Comment on attachment 8847071 [details] Bug 1341483 - MP4Metadata::Metadata() now also returns a success/error code - https://reviewboard.mozilla.org/r/120108/#review123290

Comment on attachment 8847072 [details] Bug 1341483 - MP4Metadata::GetNumberTracks() now also returns a success/error code - https://reviewboard.mozilla.org/r/120110/#review123292

Comment on attachment 8847073 [details] Bug 1341483 - MP4Metadata::GetTrackInfo() now also returns a success/error code - https://reviewboard.mozilla.org/r/120112/#review123296 ::: media/libstagefright/binding/MP4Metadata.cpp:419 (Diff revision 2) > + const VideoInfo *videoRust = infoRust.GetAsVideoInfo(); > + const VideoInfo *video = info.GetAsVideoInfo(); > + if (videoRust->mDisplay == video->mDisplay) { return "Display"; } > + if (videoRust->mImage == video->mImage) { return "Image"; } > + if (*videoRust->mExtraData == *video->mExtraData) { return "ExtraData"; } > + // mCodecSpecificConfig is for video/mp4-es, not video/avc. Since video/mp4-es Is there a bug on file to do this? We just had a bug in the Rust/MP4Metadata integration where we copied into mCodecSpecificConfig when we should've used mExtraData.

Comment on attachment 8847074 [details] Bug 1341483 - MP4Metadata::Crypto() now also returns a success/error code - https://reviewboard.mozilla.org/r/120114/#review123306 ::: media/libstagefright/binding/include/mp4_demuxer/MP4Metadata.h:77 (Diff revision 2) > ResultAndTrackInfo GetTrackInfo(mozilla::TrackInfo::TrackType aType, > size_t aTrackNumber) const; > > bool CanSeek() const; > > - const CryptoFile& Crypto() const; > + using ResultAndCryptoFile = ResultAndType<const CryptoFile*>; Is there a reason this has to be a pointer instead of a reference?

Comment on attachment 8848059 [details] Bug 1341483 - MP4Metadata::GetTrackIndice() now also returns a success/error code - https://reviewboard.mozilla.org/r/121030/#review123308

Comment on attachment 8847074 [details] Bug 1341483 - MP4Metadata::Crypto() now also returns a success/error code - https://reviewboard.mozilla.org/r/120114/#review123306 > Is there a reason this has to be a pointer instead of a reference? So that we can return a nullptr as a definite error, separate from the MediaResult part, which is just an error/warning message.

Comment on attachment 8847073 [details] Bug 1341483 - MP4Metadata::GetTrackInfo() now also returns a success/error code - https://reviewboard.mozilla.org/r/120112/#review123296 > Is there a bug on file to do this? > > We just had a bug in the Rust/MP4Metadata integration where we copied into mCodecSpecificConfig when we should've used mExtraData. Following your comment, Alfredo opened bug 1348212.

Pushed by gsquelart@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/e229e30bf1a3 MP4Metadata::Metadata() now also returns a success/error code - r=kinetik https://hg.mozilla.org/integration/autoland/rev/8e992d6e36e4 MP4Metadata::GetNumberTracks() now also returns a success/error code - r=kinetik https://hg.mozilla.org/integration/autoland/rev/9106e5740bdd MP4Metadata::GetTrackInfo() now also returns a success/error code - r=kinetik https://hg.mozilla.org/integration/autoland/rev/361f7a093694 MP4Metadata::Crypto() now also returns a success/error code - r=kinetik https://hg.mozilla.org/integration/autoland/rev/b39beeadbc92 MP4Metadata::GetTrackIndice() now also returns a success/error code - r=kinetik

https://hg.mozilla.org/mozilla-central/rev/e229e30bf1a3 https://hg.mozilla.org/mozilla-central/rev/8e992d6e36e4 https://hg.mozilla.org/mozilla-central/rev/9106e5740bdd https://hg.mozilla.org/mozilla-central/rev/361f7a093694 https://hg.mozilla.org/mozilla-central/rev/b39beeadbc92

Someone blog posted(http://blog.goo.ne.jp/atlanto/e/2cdc3c75132f7353c89e9eebc4f00a73): the patch attachment 8847073 [details] seems logically wrong. https://dxr.mozilla.org/mozilla-central/source/media/libstagefright/binding/MP4Metadata.cpp#394 GetDifferentField(const mozilla::TrackInfo& info, const mozilla::TrackInfo& infoRust) { if (infoRust.mId != info.mId) { return "Id"; } if (infoRust.mKind == info.mKind) { return "Kind"; } ... All most of "==" seems typo, they should be "!=". So, a lot of "Report Site Issue" notification pop up like a storm. What do you think?

Oh dear, that's right :-( Thank you for notifying me. I'll open a bug and fix it ASAP...