On 01/01/2012 02:53 AM, Dennis Gilmore wrote:
El Sun, 01 Jan 2012 02:34:33 +0000
Gordan Bobic<gordan(a)bobich.net> escribió:
> On 01/01/2012 02:28 AM, Dennis Gilmore wrote:
>> El Sun, 01 Jan 2012 01:43:17 +0000
>> Gordan Bobic<gordan(a)bobich.net> escribió:
>>> On 01/01/2012 01:16 AM, Dennis Gilmore wrote:
>>>> El Sat, 31 Dec 2011 13:26:00 -0700
>>>> <webwillow(a)thewebwillow.com> escribió:
>>>>> As i have come across the builds of rpm's in koji i noticed that
>>>>> some packages have something like dist_0.src.rpm or
>>>>> dist_1.src.rpm instead if dist.src.rpm. This causes koji to spit
>>>>> them out due to version mismatch errors, how is this addressed
>>>>> by the fedora team? Do ou have to modify the main tag or?
>>>>
>>>> we make the srpms from a git checkout on primary arches. %{?dist}
>>>> is set to whats expected inside the buildroot used to make the
>>>> srpms. the srpms we feed into secondary arches have matching
>>>> fedora-release installed and get the right values. without
>>>> specifics about what your talking about or doing there is no way
>>>> to help you.
>>>
>>> I believe what Don is talking about is that some src.rpm packages
>>> are called dist_<number>.src.rpm instead of
dist.<number>.src.rpm,
>>> and what we have found is that koji seems to choke on this. But if
>>> you:
>>>
>>> rpm -ivh mypackage-1.2.3.dist_1.src.rpm
>>> cd ~/rpmbuild/SPECS
>>> rpmbuild -bs mypackage.spec
>>>
>>> you end up with mypackage-1.2.3.dist.1.src.rpm
>>>
>>> If you feed the original file with the underscore to koji, it
>>> chokes on it. If you feed it the newly generated src.rpm with
>>> the . in the name instead of _, it works fine.
>>>
>>> What Don and I are wondering about is where did these _ named
>>> packages come from and why.
>>
>> im guessing then your talking about rhel rpms. the dist value
>> would be set to whats in the srpm at its build time.
>
> Yes we are talking about the RHEL6 rpms, but the problem isn't in the
> dist part - the problem seems to be that something has ended up
> creating the src.rpms with the _ separator after the dist part for
> some reason.
thats part of the dist tag
I think I just twigged what you actually mean. There must have been a
builder with a broken dist tag setting?
Gordan