On Sat, 25 Jun 2011 09:45:58 -0300, david@tethera.net wrote:
> Here is an updated version.

Thanks. I've committed this now, (waiting to be pushed until I fix my
build---the symbols stuff---so I can actually run "make test" again).

> I'm not sure the best way to do a test of
> the cleaning; maybe we should ship a MANIFEST file containing the
> output of git ls-files. I'm not sure how much churn this would cause
> in git.  Perhaps it could be treated like version, and generated from
> git if possible.

Definitely wouldn't want a generated file under revision control. It
would be possible to just run "git ls-files" if we're in a git
repository and skip the test otherwise. Or...

> In any case I guess this couldn't really be part of
> our regular test suite, because all the other tests would fail ;).

Heh, that would be a bad failure mode. :-)

The typical way to do this (as far as I understand) is as part of "make
distcheck". The idea there is to make a tar file; verify that the tar
file can be unpacked, configured, built, and installed; and then to
verify that "make distclean" returns the directory to the same state as
just after having unpacked the tar file.

I'm not really all that concerned about it though. We don't add/remove
files all that often, and it's generally not too bad of a failure mode
if "make clean" isn't perfect.

So if someone added testing for this, that would be fine, but it's not a
high priority for me.

-Carl

-- 
carl.d.worth@intel.com