This may be a trivial question, but my friend Google is not showing me any
obvious answers, so I will ask here at my own peril.
Say one needs to configure and build the same source with two (or more)
different sets of options that generate different binary RPMs, but which have
files in exactly
PS. I am aware of the alternatives approach, but looking to see whether there
is something that rpm specs have natively for this.
--
Bojan
4 Nov 2022 7:31:14 am Bojan Smojver :
This may be a trivial question, but my friend Google is not showing me any
Hello, Bojan.
The only way I know is to utilise the "RemovePathPostfixes" tag for this, e.g:
> %package gtk3
> RemovePathPostfixes: .gtk3
>
> %package qt
> RemovePathPostfixes: .qt
>
> %files gtk3
> %{_bindir}/%{name}.gtk3
>
> %files qt5
> %{_bindir}/%{name}.qt5
Then, you just need to rename th
Cool, thank you! I think this is exactly what I was looking for
(unsuccessfully).
--
Bojan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org
Fedora Code of Conduct:
https://docs.
* Bojan Smojver via devel:
> Sure, it is easy enough to configure/build repeatedly and stash the
> results into non-conflicting paths of buildroot, but how does one then
> package this in %files sections into exactly the same paths?
See tests/data/SPECS/test-subpackages-pathpostfixes.spec in the
Thank you!
--
Bojan
4 Nov 2022 8:48:25 am Florian Weimer :
* Bojan Smojver via devel:
> Sure, it is easy enough to configure/build repeatedly and stash the
> results into non-conflicting paths of buildroot, but how does one then
> package this in %files
Collections can also help manage this:
https://www.softwarecollections.org/en/
On 11/4/22 01:21, Bojan Smojver via devel wrote:
Thank you!
--
Bojan
4 Nov 2022 8:48:25 am Florian Weimer :
* Bojan Smojver via devel:
Sure, it is easy enough to configur
On 03/11/2022 22:14, Bojan Smojver via devel wrote:
PS. I am aware of the alternatives approach, but looking to see whether
there is something that rpm specs have natively for this.
You can't use alternatives on immutable Fedora versions.
--
Sincerely,
Vitaly Zaitsev (vit...@easycoding.org)
On 03/11/2022 21:31, Bojan Smojver via devel wrote:
Say one needs to configure and build the same source with two (or more)
different sets of options that generate different binary RPMs, but which
have files in exactly the same place. This is to support different
hardware. The end result would
Hi,
On 11/3/22 21:31, Bojan Smojver via devel wrote:
> This may be a trivial question, but my friend Google is not showing me any
> obvious answers, so I will ask here at my own peril.
>
> Say one needs to configure and build the same source with two (or more)
> different sets of options that g
Hi Again,
On 11/3/22 21:31, Bojan Smojver via devel wrote:
> This may be a trivial question, but my friend Google is not showing me any
> obvious answers, so I will ask here at my own peril.
>
> Say one needs to configure and build the same source with two (or more)
> different sets of options
It was for packaging xorgxrdp with glamor support. Submitted to bodhi now, so
all good.
--
Bojan
4 Nov 2022 7:38:17 pm Hans de Goede :
Hi,
On 11/3/22 21:31, Bojan Smojver via devel wrote:
> This may be a trivial question, but my friend Google is not sh
binaries from the same source
Date: 04/11/22 19:45:43
Maybe the rpath tricks + making the actual lib (or binary) a link to /run/foo
and then have a udev rule create /run/foo to point to the right version
depending on the hw it is running on might be useful for you too
I think the need to find such attempts is a clear indication there is
something wrong with the design of current implementation.
If there are binaries with different build results, I think some code
should be refactored out of the binary itself. The common parts can
remain, but hardware specif
On Fri, Nov 04, 2022 at 01:44:41PM +0100, Petr Menšík wrote:
> If there are binaries with different build results, I think some
> code should be refactored out of the binary itself. The common parts
> can remain, but hardware specific parts should be moved to
> dynamically loaded *.so files. The co
On Tuesday, 8 November 2022 09:43:40 GMT Richard W.M. Jones wrote:
> On Fri, Nov 04, 2022 at 01:44:41PM +0100, Petr Menšík wrote:
>
> > If there are binaries with different build results, I think some
> > code should be refactored out of the binary itself. The common parts
> > can remain, but hard
16 matches
Mail list logo