On Tue, 2007-10-09 at 20:52 -0400, seth vidal wrote:
> On Tue, 2007-10-09 at 20:43 -0400, Jeremy Katz wrote:
> > foo.noarch won't obsolete bar.i386 only if there's also a foo.i386.  Now
> > for dinner.
> 
> so isn't this a non-starter, then? If we don't let noarch ->arch and
> arch->noarch obsoletes then we'll be regressing vs current behavior.

 I think Jeremy's explanation had too many negatives:

1. foo.i386   obsoletes bar.noarch == works
2. foo.noarch obsoletes bar.i386   == works
3. foo.i386 and/or foo.noarch obsoletes bar.i386 == only installs foo.i386

...we don't ever do the later, do we?

 This also leads into a similar problem I've seen, my unit test is at
the end (I'm showing the test last, as I'm not 100% sure I've modeled it
properly).
 The problem being that if you upgrade X.i386 and Y is a dep. then only
Y.i386 is upgraded and not Y.x86_64 ... this then causes problems due to
file md5sum mismatches between the different versions of Y.i386 and
Y.x86_64.

 Here is the unit test that fails:

    def testInstallSinglePackageRequireUpXtra2_32MultiLib(self):
        po = FakePackage('zsh', '1', '1', None, 'i386')
        po.addRequires('zip', 'GT', ('0', '2', '3'))
        self.tsInfo.addInstall(po)

        ipo = FakePackage('zip', '1', '3', '0', 'i386')
        self.rpmdb.addPackage(ipo)
        
        xpo = FakePackage('zip', '2', '4', '0', 'x86_64')
        self.xsack.addPackage(xpo)
        xpo = FakePackage('zip', '2', '4', '0', 'i386')
        self.xsack.addPackage(xpo)

        self.assertEquals('ok', self.resolveCode())

        txmbrs = self.tsInfo.matchNaevr('zip')
        self.assertEquals(2, len(txmbrs))
        self.assertEquals(('zip', 'i386',   '0', '2', '4'), txmbrs[0].pkgtup)
        self.assertEquals(('zip', 'x86_64', '0', '2', '4'), txmbrs[1].pkgtup)


-- 
James Antill <[EMAIL PROTECTED]>

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
Yum-devel mailing list
[email protected]
https://lists.dulug.duke.edu/mailman/listinfo/yum-devel

Reply via email to