[MERGE] readdir, take 2.

[John Arbash Meinel]

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

Martin Pool wrote:
...

> 
> I'm not sure what the other pyrex tests do, but this definitely seems
> like a case where we should be indicating that some tests were not run.
> 

So, for the original pyrex tests (like dirstate, etc) that use subclassing
instead of adaptation, you just get a bunch of Feature failures for all those
tests.

Robert also used adaptation in the btree code, and my method for that was to
write *one* test that just strictly depended on the feature being available.

I'm not sure which I prefer (1 failing test focused on the feature being
available, or N tests revealing how much is actually missing.)

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIvS4pJdeBCYSNAAMRAopAAKDDrHJ4lf+gIT15jNaVI5NKi1nKqACfVPgw
b5Ru/7XzgenQB9K+3OuVOAE=
=A7AQ
-----END PGP SIGNATURE-----