Instead of using the custom 'rpm' target of setup.py,
generate a dist archive using the 'sdist' target and then
call rpmbuild ourselves: this way we can define _topdir
and stop artifacts from ending up in ~/rpmbuild.

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
---
 jobs/python-distutils.yaml | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/jobs/python-distutils.yaml b/jobs/python-distutils.yaml
index 0b20b17..5601d26 100644
--- a/jobs/python-distutils.yaml
+++ b/jobs/python-distutils.yaml
@@ -122,7 +122,9 @@
           {global_env}
           {local_env}
           sed -i -e 's/BuildRequires: libvirt.*devel.*//' *.spec.in
-          $PYTHON ./setup.py rpm
+          rm -f dist/*.tar.{{ archive_format }}
+          $PYTHON ./setup.py sdist
+          rpmbuild --clean --define "_topdir `pwd`/rpmbuild" -ta dist/*.tar.{{ archive_format }}
     publishers:
       - email:
           recipients: '{obj:spam}'
-- 
2.17.1

--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list

On Fri, Aug 10, 2018 at 03:37:41PM +0200, Andrea Bolognani wrote:
> Instead of using the custom 'rpm' target of setup.py,
> generate a dist archive using the 'sdist' target and then
> call rpmbuild ourselves: this way we can define _topdir
> and stop artifacts from ending up in ~/rpmbuild.
>
> Signed-off-by: Andrea Bolognani <abologna@redhat.com>
> ---
>  jobs/python-distutils.yaml | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/jobs/python-distutils.yaml b/jobs/python-distutils.yaml
> index 0b20b17..5601d26 100644
> --- a/jobs/python-distutils.yaml
> +++ b/jobs/python-distutils.yaml
> @@ -122,7 +122,9 @@
>            {global_env}
>            {local_env}
>            sed -i -e 's/BuildRequires: libvirt.*devel.*//' *.spec.in
> -          $PYTHON ./setup.py rpm
> +          rm -f dist/*.tar.{{ archive_format }}
> +          $PYTHON ./setup.py sdist
> +          rpmbuild --clean --define "_topdir `pwd`/rpmbuild" -ta dist/*.tar.{{ archive_format }}

So what if you used a standard bdist_rpm command from distutils core, I believe
$PYTHON ./setup.py bdist_rpm --bdist-base <foo> would be equal to your _topdir.
Although, that's just what I've digested from distutils docs, so even though
bdist_rpm has a plethora of options you can specify there can always be one
we'll be missing :P

Erik

>      publishers:
>        - email:
>            recipients: '{obj:spam}'
> --
> 2.17.1
>
> --
> libvir-list mailing list
> libvir-list@redhat.com
> https://www.redhat.com/mailman/listinfo/libvir-list

--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list

On Mon, 2018-08-20 at 16:01 +0200, Erik Skultety wrote:
> On Fri, Aug 10, 2018 at 03:37:41PM +0200, Andrea Bolognani wrote:
> > -          $PYTHON ./setup.py rpm
> > +          rm -f dist/*.tar.{{ archive_format }}
> > +          $PYTHON ./setup.py sdist
> > +          rpmbuild --clean --define "_topdir `pwd`/rpmbuild" -ta dist/*.tar.{{ archive_format }}
> 
> So what if you used a standard bdist_rpm command from distutils core, I believe
> $PYTHON ./setup.py bdist_rpm --bdist-base <foo> would be equal to your _topdir.
> Although, that's just what I've digested from distutils docs, so even though
> bdist_rpm has a plethora of options you can specify there can always be one
> we'll be missing :P

I haven't been able to find any bdist_rpm documentation that is not
filed under Python 2, which leads me to believe it might not be as
supported (if at all) under Python 3; moreover, the current
documentation[1] seems to point to FPM as the preferred way to
generate RPM packages, but that process doesn't looks like it
involves spec files at all and bundle a whole lot of other stuff
along with your actual software, so I'd say it's not really suitable
for our purpose.

In any case, I would still prefer the two-step approach (dist plus
rpmbuild) to building RPMs because it is consistent with what we do
for all other build systems (autotools and Perl's Module::Build).


[1] https://packaging.python.org/overview/#operating-system-packages
-- 
Andrea Bolognani / Red Hat / Virtualization

--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list

On Mon, Aug 20, 2018 at 04:37:13PM +0200, Andrea Bolognani wrote:
> On Mon, 2018-08-20 at 16:01 +0200, Erik Skultety wrote:
> > On Fri, Aug 10, 2018 at 03:37:41PM +0200, Andrea Bolognani wrote:
> > > -          $PYTHON ./setup.py rpm
> > > +          rm -f dist/*.tar.{{ archive_format }}
> > > +          $PYTHON ./setup.py sdist
> > > +          rpmbuild --clean --define "_topdir `pwd`/rpmbuild" -ta dist/*.tar.{{ archive_format }}
> >
> > So what if you used a standard bdist_rpm command from distutils core, I believe
> > $PYTHON ./setup.py bdist_rpm --bdist-base <foo> would be equal to your _topdir.
> > Although, that's just what I've digested from distutils docs, so even though
> > bdist_rpm has a plethora of options you can specify there can always be one
> > we'll be missing :P
>
> I haven't been able to find any bdist_rpm documentation that is not
> filed under Python 2, which leads me to believe it might not be as
> supported (if at all) under Python 3; moreover, the current
> documentation[1] seems to point to FPM as the preferred way to
> generate RPM packages, but that process doesn't looks like it
> involves spec files at all and bundle a whole lot of other stuff
> along with your actual software, so I'd say it's not really suitable
> for our purpose.
>
> In any case, I would still prefer the two-step approach (dist plus
> rpmbuild) to building RPMs because it is consistent with what we do
> for all other build systems (autotools and Perl's Module::Build).
>
>
> [1] https://packaging.python.org/overview/#operating-system-packages

Fair enough,
Reviewed-by: Erik Skultety <eskultet@redhat.com>

--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list