Hi Alan,

On Thu, 2011-05-19 at 09:07 +1200, Tim Foster wrote:
> > This doesn't seem to make sense to me, because only one variant is listed.
> 
> Yeah, what it's saying is that the variant overlaps between what's being
> linted, and the reference version - it sounds like it's not seeing the
> linted package as a successor of a reference repository, which seems
> wrong.
> 
> Could you point me (offline) to the repositories involved?

So I took a look at these, and pkglint ran without errors on those
repositories, running the version of pkglint from snv_165.


Given the package involved here, I wonder are you hitting:

18034 pkglint should cope with actions that move between packages

which was fixed in snv_164.

Part of that bug fix involved work on how we deal with package
versioning inside pkglint.  In particular, it looks like the versioning
of those packages is quite odd:

In the reference repository we have:

system%2Fmanagement%2Fipmitool/1.8.10%2C5.11-0.165%3A20110510T165929Z

and in the lint image, we have:

system%2Fmanagement%2Fipmitool/0.5.11%2C5.11-0.166%3A20110511T055754Z

I bet the old pkglint code wasn't seeing what was being delivered by the
lint image as being a successor of the one in the reference image, hence
the rather weird error message.

That said, it looks like there's vestiges of the problem remaining,
since with the fixed pkglint code, we still get:

WARNING pkglint.manifest004       last name component ipmitool in
package name clashes across
pkg://on-nightly/system/management/[email protected],5.11-0.166:20110511T055754Z 
pkg://solaris/system/management/[email protected],5.11-0.165:20110510T165929Z

which is still a bit weird, since:

>>> import pkg.fmri
>>> f1 = 
>>> pkg.fmri.PkgFmri('pkg://system/management/[email protected],5.11-0.165:20110510T165929Z')
>>> f2 = 
>>> pkg.fmri.PkgFmri('pkg://system/management/[email protected],5.11-0.166:20110511T055754Z')
>>> f1.is_successor(f2)
True

whereas pkglint's successor check disagrees, and is wrong, I think.

>>> engine.lint_fmri_successor(f2, f1)
False

        cheers,
                        tim

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

Reply via email to