[MERGE] FormatRegistry

[Aaron Bentley]

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

John Arbash Meinel wrote:
>>>register_lazy freaks me, because I can tell it's not import.  Which
>>>means I have to learn something else that's not import.  So considering
>>>how little it does, I think using register_lazy is a net loss.
> 
> 
> 
> Well, IMO it is much cleaner looking to use register_lazy.

Maybe, but I can read a helper more easily than I can read a
register_lazy_transport call.  register_lazy_transport subverts my
syntax highlighting.  And it's worse than plain register_lazy, because
it seems there's a new one for every Registry.  So even if I know what
register_transport does and I know what register_lazy does, I still have
to figure out what register_lazy_transport does.

> v- I thought the whole point of passing "default" was because you wanted
> it to explicitly exist in the list. It seems you are changing it here,
> which means that if something changes the registered "default", it still
> won't use it.

I'm hoping this will be a temporary problem.

There's no legitimate reason to change the default.  The code that does
is old code that is just trying to create a particular format, something
 that is better-handled by doing TestCase.make_branch_and_tree and
supplying a format parameter.

With the registry in place, I can update the old code, then deprecate
Repository.set_default_format.

> So other than wondering why you are hard-coding "default" +1.

Does that make sense?

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFFiwFo0F+nu1YWqI0RAhbQAJ4s6bp5MQr0K1eHfJwILhaeLlcEDACfUB4M
Jn3PhBu+fbZyPpAwepxuEDo=
=n/Hu
-----END PGP SIGNATURE-----