pungi recent bug

[Seth Vidal]

On Tue, 4 Aug 2009, Jesse Keating wrote:

> On Sat, 2009-08-01 at 03:21 +0100, Sergio Monteiro Basto wrote:
>> On Sun, 2009-07-19 at 10:57 +0100, Sergio Monteiro Basto wrote:
>>>>> That's the log I can get for wormux.
>>>>>
>>>
>>>>
>>>> So you exclude wormux but not wormux-data?
>>> yes, to just exclude wormux, I think just appears wormux in duplicates
>>> and also is enough to solve the duplicated.
>>
>> Hi, updating the information :
>>
>> I build pungi-2.0.16 reverting this commit :
>> http://git.fedorahosted.org/git/pungi.git?p=pungi.git;a=commitdiff_plain;h=44c4028447884828e01ff3769962011bc53c428d
>> and can confirm this commit is the one which breaks my "builds" .
>>
>> About seeing logs, I just found extend logs recently and if you want see
>> it too: http://sergiomb.no-ip.org/f11/logs/i386.log
>>
>> Analyzing logs:
>> wormux-data is repeated and not detect by checkduplicated because one
>> is .i586 and other is .noarch (for example in 64 bits I can have same
>> package in .i586 and in .x86_64 and they aren't duplicated)
>>
>> Pungi.INFO: Checking deps of wormux.i586
>> yum.verbose.YumBase.DEBUG: Matched wormux-data-0.8.2-5.fc11.i586 to require for wormux-data
>> yum.verbose.YumBase.DEBUG: Matched wormux-data-0.8.3-1.fc11.noarch to require for wormux-data
>> Pungi.INFO: Added wormux-data.noarch for wormux.i586
>> Pungi.INFO: Added wormux-data.i586 for wormux.i586
>>
>> this looks (to me) wrong.
>>
>> hope that can help
>> Thanks,
>
> Ah, that's interesting.  Since the arch changed, using the yum API call
> to find the best matches by arch means we'll get both matches.  I'll
> need to look into that and work with seth to find a better API call to
> use here.
>

If you add a version-specific dep for wormux-data to the wormux package 
does the problem go away?

and why do we have 2 wormux-data's in the tree?

-sv