[MERGE] Unit tests for win32 glob expansion

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Kuno Meyer wrote:
> And once again, this time hopefully without any stupid errors...
> 
> Kuno

The only thing I would make a comment on, is that we haven't been using Feature
to represent platform specific limitations.

I'm not sure what to think just yet....

Specifically, I was considering that on our PQM test bot should cause tests to
fail if they are missing a feature, which would help ensure that all tests are
being run.

I know Martin is looking into having more error classes for 'failing' tests.
We have both:

  http://bazaar-vcs.org/BzrExtendTestSuite

And his thread of "[rfc] add more possible test results".


This would seem like a TestPlatformLimit according to his email. More than
TestDependencyMissing.

I at least think of Features as mostly "DependencyMissing".

However, I guess we could extend the test infrastructure, so that a given
feature could indicate what error subclass it was.

Certainly, one of the documented possible Features was Symlinks, and that is a
Platform limitation, not a missing dependency.

So +1 from me, modulo a discussion on the list about how we want to handle
Features that cannot be fulfilled.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGnp8FJdeBCYSNAAMRAj1OAJ99fwgMDyZ3Cdq6GvEqQn6oMBfc/ACgol8N
nX/eizsW7fHlz9VVCq4HPPE=
=p+Sb
-----END PGP SIGNATURE-----