[FFmpeg-devel] [PATCH][GSoC] dshow add support for saving and loading of capture devices

[Roger Pack]

On 4/22/15, Máté Sebők <smfinc.org at gmail.com> wrote:
> Added NULL-ing after release on load and save, lower-cased the #include
> "shlwapi.h" and added release of pers_stream in the error handler.

OK the crash is gone now.
I'd like to have the IStream_Release only in one section, however.
Would it be possible to remove the Release's and *only* release in the
error block please?

> There is no other pers_stream, only in the save because the
> OleLoadFromStream loads directly to the described object.

OK point taken.  I'd like to see the IPersistStream Release only
called in the error block as well.  Just to match the existing code
style.  It's practically there, thank you!
-roger-

> On Wed, Apr 22, 2015 at 4:46 PM, Roger Pack <rogerdpack2 at gmail.com> wrote:
>
>> On 4/22/15, Michael Niedermayer <michaelni at gmx.at> wrote:
>> > On Wed, Apr 22, 2015 at 01:40:03PM +0200, Hendrik Leppkes wrote:
>> >> On Wed, Apr 22, 2015 at 1:12 PM, Michael Niedermayer
>> >> <michaelni at gmx.at>
>> >> wrote:
>> >> > On Wed, Apr 22, 2015 at 12:47:59PM +0200, Máté Sebők wrote:
>> >> >> Indeed, sorry i've forgotten to change the Eclipse' EOL to unix--
>> >> >> now
>> >> >> fixed.
>> >> >>
>> >> >> Error handler fixed.
>> >> >>
>> >> >> Parameter names changed to audio(/video)_device_load(/save) as I
>> think
>> >> >> the
>> >> >> to_file and from_file is evident and is specified in the help.
>> >> >>
>> >> >> About the "fake" input:  by loading a device filter from file, in
>> >> >> the
>> >> >> dshow_open_device() it loads the device object from file instead of
>> >> >> searching it by name. So as long as the dshow_open_device() get
>> called
>> >> >> it
>> >> >> will load the device -- however to get called it is required to
>> >> >> have
>> a
>> >> >> not
>> >> >> NULL input name .. aka any fake text would do the trick (eg: -i
>> video="
>> >> >> " )
>> >> >> If you have better suggestion for the description, please tell me.
>> >> >>
>> >> >> Here is the fixed patch.
>> >> >
>> >> > [...]
>> >> >
>> >> >> diff --git a/libavdevice/dshow.c b/libavdevice/dshow.c
>> >> >> index d03670e..9ac28c5 100644
>> >> >> --- a/libavdevice/dshow.c
>> >> >> +++ b/libavdevice/dshow.c
>> >> >> @@ -27,6 +27,8 @@
>> >> >>  #include "libavformat/riff.h"
>> >> >>  #include "avdevice.h"
>> >> >>  #include "libavcodec/raw.h"
>> >> >> +#include "objidl.h"
>> >> >
>> >> >> +#include "Shlwapi.h"
>> >> >
>> >> > this header does not exist, it must be lowwer case or it will break
>> >> > build on linux-mingw
>> >> >
>> >>
>> >> That is the proper capitalization of the header in the MS Windows SDK.
>> >> If mingw has it all lower-cased, its technically a bug in mingw.
>> >
>> > hmm i wasnt aware of that but yes
>> >  locate Shlwapi.h confirms that theres no such header in ubuntu mingw*
>> >
>> > so this definitly needs some configure check so it doesnt break build
>>
>> Interesting.  After doing some research (in my own cross compiler, I
>> only have "shlwapi.h" and not "Shlwapi.h"), it appears that the
>> mingw-gcc cross compilers are themselves "filename case insensitive"
>> even on linux with case sensitive underlying file systems.  So I guess
>> as long as "shlwapi.h" is common we should be ok.  Unexpected for
>> sure.
>> _______________________________________________
>> ffmpeg-devel mailing list
>> ffmpeg-devel at ffmpeg.org
>> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>
>