Backwards compatibility of the 'authors' revision property?

James Westby jw+debian at jameswestby.net
Sat Mar 7 20:44:53 GMT 2009


On Fri, 2009-03-06 at 20:26 +1100, Ben Finney wrote:
> Matt Nordhoff <mnordhoff at mattnordhoff.com> writes:
> 
> > (Disclaimer: I haven't actually tested this. I just glanced at the code.)
> > 
> > When the 'authors' revision property was added, ISTM it was changed so
> > that new revisions will only ever have the 'authors' revprop, and no
> > 'author' revprop. This is bad for interacting with older clients, isn't
> > it? They'll only miss a bit of information in "bzr log", but still.
> 
> Yes, that would be bad. My understanding was that there will still be
> a single-person ‘author’ property set on every revision, so that tools
> requiring the concept of “single person who owns this revision”
> still have it. Is that not the case?

That is not the case, as it wasn't entirely clear who the single person
would be. At a semantic level, if we pair on a new feature, who owns it?
At a practical level, is it clear enough that the first "--author" 
argument on the command line will be the person that is said to own the
revision?

As Matt points out in the reply, there are some places in bzr that 
aren't ready to handle multiple authors in the output, e.g. annotate,
so I took the easy way out and just showed a single author, the first
in the list. It would be great to teach the rest of bzr about multiple
authors so that this could be avoided.

Thanks,

James




More information about the bazaar mailing list