depcheck test (was Re: measuring success)

Will Woods wwoods at redhat.com
Wed Jul 7 15:51:22 UTC 2010


On Tue, 2010-07-06 at 12:11 -0700, Adam Williamson wrote:
> On Tue, 2010-07-06 at 11:34 -0400, Will Woods wrote:
> 
> > If there are any other questions, feel free to ask.
> > 
> > -w
> 
> Did you get to look at the nss-softokn situation (details of which I
> sent to autoqa-devel) yet? How hard would it be to catch that?

Yes - as I understand it, the reason the nss-softokn.i686 update didn't
land in the x86_64 updates repo was a shortcoming of mash's multilib
handling algorithm. 

Since depcheck will need to run mash on the set of proposed updates in
order to correctly handle multilib cases, it should (in theory)
correctly reproduce this error and thus set off alarm bells. 

I'm not sure how transparent the error messages will be, though, but at
least (again, in theory) we should get notification that Something Is
Wrong.

We'll see what happens when we actually have a test case for this[1],
though.

-w

[1] Real Soon Now, of course



More information about the devel mailing list