Bugfix/js functions use single pointer options #36
Conversation
The JS versions of the functions should take single-pointer AVDictionary-options. This fixes Yahweasel#34
This change only tests adding any kind of options to libav.av_dict_set_js. Without this PR this results in an error (since it will try to free something that is not a pointer), with this PR this will not result in an error anymore. Note that the option set in the test is unknown to the demuxer and has no effect.
|
I really can't thank you enough for finding these things. This is a corner of the types that I'd never really fought, so never really tested. I'll review these changes later today. By the way, I believe that in one of these earlier interactions you mentioned that you were doing something with HLS, so if you didn't notice, you may want to know that I added HLS support directly to libav.js. For the tests I've done it works great, but it's also not in the test suite, since that requires external stuff to test correctly ^^´. You can find info on it in IO.md. |
|
No worried, happy I can contribute to this amazing project :) I was not the person asking about HLS -- I'm only trying to read/write to/from FileSystemFileHandles (for now). Still, good reminder to check what is new in 4.6.6 (I like commits called "Switched to a much more robust dependencies system" :)). |
As discussed in #34 , this PR patches both
av_dict_set_jsandavio_open2_jsto accept aAVDictionary*rather than aAVDictionary **(since the JS code only exposedAVDictionary*).Note that this does mean that
av_dict_set_jswill not be able to return a list of unused options (as ffmpeg'sav_dict_setdoes).I changed the
500-demuxer-decodetest to include some arbitrary options dictionary. This test will fail without the code change, since free will be called on something that is not a pointer.