[MERGE] catch-all handler for filter stack lookup

[Jelmer Vernooij]

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

Ian Clatworthy wrote:
> Jelmer Vernooij wrote:
> 
>>> I'm making this an RFC because I want Jelmer to confirm
>>> this meets his needs and because I'm not sure how best
>>> to expand the lazy registration function Jelmer added.
>>> Perhaps we want to register a compound object rather
>>> than a (map,fallback) tuple? If so, we could point to that
>>> object. Or we could point lazily to both a map and a fallback
>>> callable?
>> Thanks! I can confirm this would meet my needs, but it would also be
>> nice to load the svn keywords support in bzr-svn lazily.
>>
>> Perhaps just the fallback parameter could be supported for lazy loading?
>> The dictionary lookup can be done inside of the fallback function as well.
> 
> That's true, but I'd prefer to keep the semantics of the fallback
> the same, regardless of how it is registered. I've therefore gone
> with (lazy) registering a (filter_stack_map, fallback) tuple and
> I've updated the docstring to reflect this. I've removed the RFC
> from this patch accordingly.
> 
> FWIW, I'm still not quite ready to make the registry used behind
> the scenes public yet. My main concern is that we'll need some
> way of optionally giving a priority to different filters so
> that things get done in a particular order. That's YAGNI currently
> but it's my reason for retaining the special registration functions
> for now.
That works for me too, thanks!

bb:approve

Cheers,

Jelmer

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iJwEAQECAAYFAknO7T0ACgkQDLQl4QYPZuVCCwP+Lk15vKgy8yzF3O7fMqbeg47t
FpEa6qrGnfpAhv0IpDV7O/XWDtRB5LC5H+RA+niUPgf3LAfhSykD5t7we/k8iEc7
cLCyE2NedSYE4b0NNC//O16SlM24qIy/pKDUcaF05NZi3jCkWmaMbhvp6yzEMqFv
csRTKDjgiJ5HXJ3H2x0=
=uFnb
-----END PGP SIGNATURE-----