[MERGE] win32-related selftest fixes

[John Arbash Meinel]

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

John Arbash Meinel wrote:
> John Arbash Meinel has voted -0.
> Status is now: Approved
> Comment:
> The tests should only be skipped on old branch formats.
> 
> If they are failing on new branch formats, we have a bug that needs to
> be fixed.
> 
> The whole *point* of this test was to make sure that we support funky
> file ids on windows with all formats newer that Weaves.
> 

Just to follow up on this a bit. There are probably some
inter-repository combinations that this test should be skipped for. (If
the source format is Weave or earlier). But we need to be a little
careful that we don't skip just because we are failing. As it may be a
sign that we are really failing.

So +1 to skipping if you provide more safety checks about when things
are going to be skipped.

Maybe we need a field on repository formats with
"supports_funky_file_ids" which defaults to True, but we could
explicitly set to False for Weave formats.

I personally would probably also be okay with an explicit
isintance(source_repo, (RepositoryFormat5, RepositoryFormat6,
                        WeaveMetaDirRepository)):

(I don't know if we also need to check target_repo)

And I should temper my comment a little bit. This specific test is to
make sure that 'file_ids_affected_by' knows to decode the XML, and that
any other such optimization we put in the future will also handle things
that must be escaped for whatever the storage format is.

The other tests is more the one in
repository_implementations/test_file_id_involved.py

Though really that one should have even more precautions about not
skipping with newer repository formats.

John
=:->



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFw5dqJdeBCYSNAAMRAr1ZAKCcbYLPYIkPsvuhEWySxwVe0+4E2ACfZBu1
o9neSovMixzKQM2suHAnjFU=
=Dnwm
-----END PGP SIGNATURE-----