[MERGE][0.90][Bug 133965] PathNotChild, port mismatch with "bzr info" for bzr:// smartserver

Robert Collins robertc at robertcollins.net
Mon Aug 27 08:16:03 BST 2007


On Mon, 2007-08-27 at 17:10 +1000, Andrew Bennetts wrote:
> 
> > John's fix is the right one I think, it's coherent with how the
> > default port is handled by other schemes.
> 
> Well, I think my fix is better for bzr.dev.  I can see the case for
> John's fix
> for 0.90 instead of mine (it's simpler and thus safer), but I'll leave
> the
> release manager to make that judgement.

I think having a different fix in 0.90 would make backporting other
fixes harder. -1 on that.

> >     john> Arguably the _split_url and _unsplit_url could be moved
> >     john> onto the base Transport class rather than the derived
> >     john> ConnectedTransport. I don't remember exactly why
> >     john> Vincent wanted the split, but I think there was a
> >     john> reason.
> > 
> > Host, port and credentials are needed for connected transports only,
> > but Transport can have a simpler definition for split and unsplit
> > if needed.
> 
> Well, the URL syntax isn't going to vary for us by scheme is it?  Some
> parts
> might not apply to all protocols (e.g. there's no user@ for bzr://,
> and no host
> or port for file://), but that doesn't change the parsing AFAICS.  So
> I think
> split/unsplit should global functions, and it's up the individual
> transports to
> explicitly raise errors on obviously impossible urls (e.g.
> MemoryTransport could
> assert that host is None).

I agree with this.

-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: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070827/5e93fd10/attachment.pgp 


More information about the bazaar mailing list