[FFmpeg-devel] [PATCH 1/2] Allow overriding SAMPLES by environment variable

Reimar Döffinger Reimar.Doeffinger
Thu Feb 10 20:21:08 CET 2011


On Wed, Feb 09, 2011 at 11:57:43PM +0100, Diego Biurrun wrote:
> On Wed, Feb 09, 2011 at 11:32:38PM +0100, Janne Grunau wrote:
> > On Wed, Feb 09, 2011 at 02:19:17PM -0800, Baptiste Coudurier wrote:
> > > 
> > > On 02/09/2011 02:13 PM, Diego Biurrun wrote:
> > > > On Wed, Feb 09, 2011 at 07:32:06PM +0100, Reimar D?ffinger wrote:
> > > >> ---
> > > >>   configure |    2 +-
> > > >>   1 files changed, 1 insertions(+), 1 deletions(-)
> > > >>
> > > >> --- a/configure
> > > >> +++ b/configure
> > > >> @@ -3248,7 +3248,7 @@ SLIB_CREATE_DEF_CMD=${SLIB_CREATE_DEF_CMD}
> > > >>   SLIB_UNINSTALL_EXTRA_CMD=${SLIB_UNINSTALL_EXTRA_CMD}
> > > >> -SAMPLES=$samples
> > > >> +SAMPLES?=$samples
> > > >
> > > > Is setting it through configure a hassle?
> > > 
> > > I think so.
> > 
> > make SAMPLES=path/to/fate
> > 
> > works without specifying the path at configure time
> 
> I thought I had mentioned this at some point already, apparently
> I did not.  Is that enough for you, Reimar?

No. There is no one big issue I am trying to solve, but a whole
bunch of minor annoyances:
specifying it for configure means you either have to use a wrapper
script or write the path each time. It also means you have
to rerun configure when you move the samples.
Specifying it on the make command-line also means writing it each
time or using a wrapper.
For both, forgetting it means that "make fate" (I think at least
it was that command) silently does not run those tests.
Allowing it to be set via an environment variable allows fixing
all these minor issues in a very convenient way by just setting
it up once in the login script and it works even in case you make
a completely new checkout or whatever.
Note that for this purpose it would be completely sufficent (maybe
even preferable) if the environment variable was used only
when a samples path is not specified in any other way.



More information about the ffmpeg-devel mailing list