qt3.3-devel providing qt-devel ?

Jonathan Underwood jonathan.underwood at gmail.com
Sat May 5 21:15:43 UTC 2012


On 5 May 2012 21:53, Paul Howarth <paul at city-fan.org> wrote:
> On Sat, 5 May 2012 21:32:49 +0100
> Jonathan Underwood <jonathan.underwood at gmail.com> wrote:
>
>> Hi,
>>
>> I am trying to build djview4 for el6. The spec file has a
>> BuildRequires: qt-devel >= 4.2. However, for some reason during
>> package building qt3.3-devel is pulled in (both on the builders and in
>> mock). I notice that:
>>
>> $ rpm -q --provides qt3-devel
>> ...
>> qt-devel = 1:3.3.8b-30.el6
>> qt3-devel = 3.3.8b-30.el6
>> ...
>>
>> which seems slightly weird, but I'd still not expect that to satisfy
>> the BuildRequires for qt-devel, since I'm requiring >= 4.2.
>
> 1:3.3.8b > 0:4.2, so the requirement is satisfied.
>

Aha, mystery solved, thanks. So, it was the epoch.

>> Looking at qt4-devel I see it provides (amongst other things):
>>
>> qt4-devel = 4.6.2-20.el6
>> qt-devel = 1:4.6.2-20.el6
>>
>> So, why isn't this being pulled in? Is it because of the epoch?
>
> Given that both packages satisfy the requirement, I'd bet yum is
> picking qt-devel rather than qt4-devel by virtue of its shorter name.
>
>> And
>> why isn't the build simply failing at satisfying the BuildRequires?
>> (It fails later on during %build for related reasons). Is this a bug
>> with the qt3.3 package?
>
> BuildRequire: qt4-devel >= 4.2
>
> should do what you want.

Yep, that's what I did, thanks.

I tend to think that qt3-devel providing qt-devel is a bit of a
packaging bug. But, I suppose requiring qt4-devel is safer and the
correct thing to do in any case.

J.




More information about the epel-devel-list mailing list