[MERGE] LockDir.lock_write() shouldn't require an email address

[Richard Wilbur]

On Fri, 2006-09-29 at 15:26 -0500, John Arbash Meinel wrote:
> The attached patch fixes a small bug in LockDir.
> 
> Basically, it requires a semi-valid email address to be configured, or
> '.lock_write()' will fail.
> 
> Also, extract_email_address() shouldn't be raising a generic BzrError.
> So this patch updates it to raise a better exception, and then has the
> callers catch the right exception.
> 
> Because of inheritance, we don't need to do a deprecation phase, because
> NoEmailInUsername is a child of BzrError.
> 
> John
> =:->
> 
John,

Did this make it into version 0.11?  I didn't see anyone else vote for
or against it.

I am in favor of the more descriptive error class as it makes the code
easier to read.  I also like the new fallback for lockDir.lock_write().
It seems like a reasonable choice to use the username when email address
is not configured as this is just to identify who current owns the lock
on a particular directory, right?

Mi dos centavos:  +1 (for what it's worth).

Richard Wilbur