Ronald S. Bultje <[email protected]> added the comment: > Ronald, what's your opinion on this?
If they're not using _open(), then they shouldn't use _close(). If they want to use any _close(), then they need to use _open(), as you suggested. > av_global_network_init() I don't think that's a good idea. The current system works just fine unless you start doing funky/fuzzy stuff like (apparently) XMBC is doing. The problem with all of these is that FFmpeg becomes less a convenience library, and more a direct wrapper of the common least denominator of set(win32,linux,macosx), and that's not right. We happen to handle this specific exception where Windows is different from Linux/Mac, and we do a good job. I don't think our API should reflect specific Win32 defects publically if there isn't a really good reason. So what's the problem with using url_open_protocol() with a custom-made protocol (xmbc://, if you want)? That'll fix all of this, right? ________________________________________________ FFmpeg issue tracker <[email protected]> <https://roundup.ffmpeg.org/issue1849> ________________________________________________
