Summary/Minutes from today's FESCo Meeting (2014-01-15)

Miroslav Suchý msuchy at redhat.com
Mon Jan 20 08:48:05 UTC 2014


On 01/16/2014 08:12 PM, Adam Williamson wrote:
> So, in the discussion of this, the following was presented as an
> obstacle:
> 
> 18:31:13 <notting> Requires: foo > 1.0-1.%{release}
> 18:31:22 <notting> 1.0-1.fc20.copr *satisfies* that


I have no objections against disttag. Hoever I would prefer "fc20-copr". I.e. not to use dot as separator.

> Well, sure. But as mitr noted in passing - and everyone seemed to ignore
> - that's a terrible conditional in all sorts of ways:
> foo-1.0-1.%(nextrelease) satisfies that conditional too, even though
> it's probably identical to foo-1.0.1.%{release} .
> 
> Does anyone have a case where %{release}.copr can cause problems that
> can't *also* be caused just by the same build having been done for
> multiple %{release}s?
> 
> In the cited case, I'd use:
> 
> Requires: foo >= 1.0-2
> 
> or something similar. In general, isn't it pretty much universally
> accepted that you should try *really hard* to avoid the disttag being
> significant to your conditionals because it's just fundamentally
> unreliable to use it?

Well there is no solution. IMHO. If you would come with solution which is smaller then original disttag you will have
problems with evaluation Obsoletes.
So you will either one problem or other.
Let the maintainer of that package in Copr projects solve it.

-- 
Miroslav Suchy, RHCE, RHCDS
Red Hat, Software Engineer, #brno, #devexp, #fedora-buildsys


More information about the devel mailing list