[MERGE] Speedup selftest by loading a previously saved test list

[John Arbash Meinel]

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

Vincent Ladeuil wrote:
>>>>>> "john" == John Arbash Meinel <john at arbash-meinel.com> writes:
> 
> <snip/>
> 
>     john> BB:comment
> 
>     john> Overall this looks like a decent patch. I wonder if we shouldn't
>     john> continue to use test_suite() when possible and just have it thunk into
>     john> load_tests(...).
> 
> We can't. We want to share the loader and using test_suite (which
> takes no parameter) makes it impossible.
> 
>     john> I'm not sure if it is worthwhile as long as we keep
>     john> bzrlib.tests.test_suite().
> 
>     john> I also wonder how broken we are relative to other test
>     john> suite runners,
> 
> Care to mention some ? But keep in mind, that this patch change
> only the way we *load* the tests, not how we run them.
> 
>     john> 
>     john> but I don't see any direct indication of our test suite
>     john> not working for it, since we still have our topmost
>     john> .test_suite() call which is still returning TestSuite
>     john> objects, etc. So I don't think your patch would change
>     john> our compatibility in an direct sense.
> 
> Especially since test_suite() looks like a bzrism AFAIK.
> 
>            Vincent
> 

.test_suite() is a Unittest ism.

Which is why we keep it so that the other python TestRunners can still
run our test suite. Which is why I was considering leaving them in other
locations.

John
=:->

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

iD8DBQFHzYwkJdeBCYSNAAMRAk+dAKCiEZ+eRgYK0dhH/R2OP1NEJtDJ2gCfYmiq
kT4NaXNZyH5BT+aNVDuQm2A=
=nqGj
-----END PGP SIGNATURE-----