Hi Shawn,

On Wed, 2010-02-24 at 15:37 -0600, Shawn Walker wrote:
> On 02/24/10 09:06 AM, Tim Foster wrote:

> > I think I had been trying to see if there was a better exception I could
> > raise for this case, where we have a valid URL that we just can't
> > support, as opposed to an invalid URL

> Really, in the end, an unsupported URI is the same as an invalid one to 
> the client.  If we need to make the distinction later on, we can have a 
> different transaction exception or add a property to the existing exception.

Yep, no problem.

> > I've updated the webrev, adding a test I should also have included:
> > checking that pkgsend create-repository returns 0 isn't enough, we need
> > to also make sure that the repository we think we created actually
> > exists each time we try a different URL.

> tests/cli/t_pkgsend.py:
>    lines 253-254: This seems a bit confusing as the pkgsend 
> create-repository step is what created the repository, not starting the 
> depot in readonly mode.  If you're actually testing the depot, you need 
> to call rmtree() before starting it on the repo pkgsend created.

Right - that's what I'm doing on line 249.  I'm using this check, just
as the old version of code was, as a way to ensure the repository we
created was acceptable to the depot server.

I've left the webrev unchanged at
http://cr.opensolaris.org/~timf/14022-pkgsend-fileurl/

        cheers,
                        tim

_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to