On Wed, 12 Jul 2006 08:02:46 -0400, Jesse Keating wrote:
On Wednesday 12 July 2006 06:14, Michael Schwendt wrote:
> Else it would be abuse of tags. The only vaguely valid case is combining
> BuildArch noarch and ExcludeArch, which is like saying "by nature, the
> package contents are arch-independent, but we know that there is a problem
> on the N excluded archs".
How far do you take this though?
ExcludeArch: ppc ppc64 s390 s390x alpha sparc sparc64 ia64 arm .....
It starts to get silly when you have to guess at the arches that somebody may
attempt to install your package for.
No, it is silly to not make the package arch-specific in that case.
Your package excludes almost a dozen archs => it is not arch-independent!
Why would 'ExcludeArch' be ok,
but 'ExclusiveArch' not be?
I tried to explain this in my previous reply.
ExcludeArch => we know about problems with the explicitly excluded archs,
we assume the package is okay for all other archs
ExclusiveArch says "I know it _only_ works here,
nowhere else."
Then it is NOT noarch.
where as ExcludeArch would be "These are the arches out in the
world I know of, and I know it doesn't work there."
Once more, ExcludeArch and ExclusiveArch are NOT information that
can be found in the .noarch package.
I agree that we are overloading the tag. But if we're going to
ban one
method, we should ban them all and make it a hard rule that if your noarch
code only works on specific arches, make it an arch specific package.
Exactly.