[MERGE] WorkingTree.unversion

[Robert Collins]

On Thu, 2006-09-07 at 17:14 -0500, John Arbash Meinel wrote:
> Robert Collins wrote:
> 
> ...
> 
> > 
> > All the others that I know of are the explicit multi-line version.
> > 
> > I dont like that abuse of kwargs to set value, for some reason I thought
> > it had been removed in BzrNewError.
> 
> I kind of liked using kwargs for it, because it made it clearer that you
> expected BzrNewError to know about a member so it could create the doc
> string later.
> 
> But I don't really care either way. Just that knowing:
> 
> self.foo = bar
> 
> Creates an entry in self.__dict__ is a little bit of voodoo.
> 
> Speaking of error voodoo. What do you think about changing the base
> error so that instead of using self.__doc__, we create a class member
> like error_str = ''

I'm very much for this. It also helps with i18n in the future. I find
the current docstring abuse rather distasteful.

> While I'm on that topic, though. I might also want to mention that for
> many of our exceptions, the user-visible value is sub-optimal. For
> example a lot of them give BzrBranch5('file://url/to/branch/that/failed)
> Rather than just:
> Could not do Foo in /path/to/branch/that/failed.
> 
> I'm thinking we probably should extend our Error api a bit. This could
> include giving an encoding for the error string, so that urls can show
> up as nice unicode paths.


-Rob
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060908/c1d44e93/attachment.pgp