On 01/23/16 11:19 AM, Hendrik Leppkes wrote:
On Sat, Jan 23, 2016 at 8:16 PM, Andreas Cadhalpun
<andreas.cadhal...@googlemail.com> wrote:
On 23.01.2016 17:16, Hendrik Leppkes wrote:
On Sat, Jan 23, 2016 at 4:27 PM, Andreas Cadhalpun
<andreas.cadhal...@googlemail.com> wrote:
On 23.01.2016 15:07, Hendrik Leppkes wrote:
On Sat, Jan 23, 2016 at 2:51 PM, Andreas Cadhalpun
<andreas.cadhal...@googlemail.com> wrote:
On 23.01.2016 14:48, Hendrik Leppkes wrote:
Unfortunately, cygpath is not guaranteed to be available, and in fact
is not on some working build setups today (ie. msys1 environments with
msvc)

Is there another way to get the Windows path on those environments?


Not that I am aware.

Hmm, sed should be available everywhere.

Does the attached patch fix the MSVC builds?


Unfortunately these assumptions are also not valid. On cygwin for
example, a path would be /cygdrive/C/path/to/folder

How many more strange path conventions are there?

I think all mentioned so far could be supported by using cygpath,
if available, and sed otherwise.

Dealing with absolute paths in the unix/windows mixture is just not
going to be easy, which is why we were careful to avoid them at all
before.

Not really. Out-of-tree builds used absolute path before, which is
exactly the problem my initial patch fixed.


For source files yes, but not for output files, which is the really
problematic case with the MSVC CLI syntax.


Does it work better without the drive delimiter? That's how I run fate, eg /fate instead of g:/fate with everything needed on G: or is Windows too anal about path separators?
Dave
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Reply via email to