On 11/29/18 11:49 PM, Andrey Semashev wrote: > On 11/29/18 2:17 PM, Andrey Semashev wrote: >> On 11/29/18 2:15 PM, Andrey Semashev wrote: >>> On 11/28/18 7:47 PM, Jeyapal, Karthick wrote: >>>> >>>> On 11/28/18 4:46 PM, Andrey Semashev wrote: >>>>> The URI used to open the output streams may be an actual URI with "file" >>>>> scheme, >>>>> according to https://tools.ietf.org/html/rfc8089. This commit makes file >>>>> deletion routine recognize file URIs and extract the actual filesystem >>>>> path >>>>> from it. >>>> There is already some code in ffmpeg to handle this. It is present in >>>> file_delete() function in file.c. >>>> We will need to avoid code duplication for the same functionality. One >>>> option could be to call avpriv_io_delete() function instead of calling >>>> unlink, so that file_delete function gets called. >>>> Calling avpriv_io_delete will also make the delete functionality easily >>>> extendable for other output protocols. >>> >>> That would be fine with me, but I'm using Linux. Looking at file_delete (in >>> libavformat/file.c), it looks like it will only work on POSIX systems but >>> not on Windows, since it doesn't have unistd.h. Am I correct? And if so, is >>> avpriv_io_delete still the preferred approach? >> >> Also, that code doesn't seem to support the URI with an authority field and >> doesn't check the special "localhost" case. > > I've sent a new set of patches that updates both file.c and dashenc.c. Thanks for your understanding. Looks like that will be the clean approach for fixing this problem. > _______________________________________________ > ffmpeg-devel mailing list > ffmpeg-devel@ffmpeg.org > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
