Hi list,
I'm updating a python package and I have the following issue. As per Fedora packaging guidelines, we should use:
%build %{__python2} setup.py build
%install %{__python2} setup.py install -O1 --skip-build --root %{buildroot}
which is fine to me. But with this new version of my package, the module doesn't get copied if --skip-build is present (it only copies the egg):
running install running install_lib warning: install_lib: 'build/lib' does not exist -- no Python modules to install ...
What should I do? If I remove --skip-build, it works fine, but I'm not sure if I can do that in the spec file.
Thanks in advance, Germán.
On Wed, Mar 12, 2014 at 05:23:17PM -0300, "Germán A. Racca" wrote:
Hi list,
I'm updating a python package and I have the following issue. As per Fedora packaging guidelines, we should use:
%build %{__python2} setup.py build
%install %{__python2} setup.py install -O1 --skip-build --root %{buildroot}
which is fine to me. But with this new version of my package, the module doesn't get copied if --skip-build is present (it only copies the egg):
running install running install_lib warning: install_lib: 'build/lib' does not exist -- no Python modules to install ...
What should I do? If I remove --skip-build, it works fine, but I'm not sure if I can do that in the spec file.
Where's the srpm at? I'll take a look.
-Toshio
On 03/12/2014 07:45 PM, Toshio Kuratomi wrote:
On Wed, Mar 12, 2014 at 05:23:17PM -0300, "Germán A. Racca" wrote:
Hi list,
I'm updating a python package and I have the following issue. As per Fedora packaging guidelines, we should use:
%build %{__python2} setup.py build
%install %{__python2} setup.py install -O1 --skip-build --root %{buildroot}
which is fine to me. But with this new version of my package, the module doesn't get copied if --skip-build is present (it only copies the egg):
running install running install_lib warning: install_lib: 'build/lib' does not exist -- no Python modules to install ...
What should I do? If I remove --skip-build, it works fine, but I'm not sure if I can do that in the spec file.
Where's the srpm at? I'll take a look.
-Toshio
Hi Toshio,
I didn't build the srpm yet, but this is the spec file so far:
http://paste.fedoraproject.org/84109/44860051/
Germán.
On Wed, Mar 12, 2014 at 08:52:25PM -0300, "Germán A. Racca" wrote:
On 03/12/2014 07:45 PM, Toshio Kuratomi wrote:
On Wed, Mar 12, 2014 at 05:23:17PM -0300, "Germán A. Racca" wrote:
Hi list,
I'm updating a python package and I have the following issue. As per Fedora packaging guidelines, we should use:
%build %{__python2} setup.py build
%install %{__python2} setup.py install -O1 --skip-build --root %{buildroot}
which is fine to me. But with this new version of my package, the module doesn't get copied if --skip-build is present (it only copies the egg):
running install running install_lib warning: install_lib: 'build/lib' does not exist -- no Python modules to install ...
What should I do? If I remove --skip-build, it works fine, but I'm not sure if I can do that in the spec file.
Where's the srpm at? I'll take a look.
-Toshio
Hi Toshio,
I didn't build the srpm yet, but this is the spec file so far:
I got busy but I started taking a look. It seems like the build scripts for that package (it's setup.py) are overriding some of the standard distutils commands with ones from astropy. I need to look at those and see if they're what's causing this issue.
-Toshio
On 03/13/2014 06:57 PM, Toshio Kuratomi wrote:
On Wed, Mar 12, 2014 at 08:52:25PM -0300, "Germán A. Racca" wrote:
On 03/12/2014 07:45 PM, Toshio Kuratomi wrote:
On Wed, Mar 12, 2014 at 05:23:17PM -0300, "Germán A. Racca" wrote:
Hi list,
I'm updating a python package and I have the following issue. As per Fedora packaging guidelines, we should use:
%build %{__python2} setup.py build
%install %{__python2} setup.py install -O1 --skip-build --root %{buildroot}
which is fine to me. But with this new version of my package, the module doesn't get copied if --skip-build is present (it only copies the egg):
running install running install_lib warning: install_lib: 'build/lib' does not exist -- no Python modules to install ...
What should I do? If I remove --skip-build, it works fine, but I'm not sure if I can do that in the spec file.
Where's the srpm at? I'll take a look.
-Toshio
Hi Toshio,
I didn't build the srpm yet, but this is the spec file so far:
I got busy but I started taking a look. It seems like the build scripts for that package (it's setup.py) are overriding some of the standard distutils commands with ones from astropy. I need to look at those and see if they're what's causing this issue.
-Toshio
Thanks for your time. If you prefer, you can contact me via IRC in Fedora channels, my nick is skytux.
Germán.
On Thu, Mar 13, 2014 at 07:03:19PM -0300, "Germán A. Racca" wrote:
On 03/13/2014 06:57 PM, Toshio Kuratomi wrote:
I got busy but I started taking a look. It seems like the build scripts for that package (it's setup.py) are overriding some of the standard distutils commands with ones from astropy. I need to look at those and see if they're what's causing this issue.
-Toshio
Thanks for your time. If you prefer, you can contact me via IRC in Fedora channels, my nick is skytux.
Okay, it looks like it is a bug in astropy.
I'm attaching a patch that I applied to astropy that got the build working. I don't have a deep understanding of setuptools and distutils internals, so I'm not sure if the patch is correct or just is a good starting point for getting the code fixed.
-Toshio
Ugh. Sorry -- the patch is essentially correct but I was too quick to think it was finished. The patch is hardcoding the build directory to "build/lib.linux-x86_64-2.7/". This is wrong. To finish off the patch and make it suitable for applying to astropy it needs to figure out that value from the build environment similar to what's being done in AstropyBuildPy.finalize_options. On different arches and systems it would be build/lib.linux-x86_64-2.7, build/lib.linux-x86-2.7, etc.
-Toshio
On Thu, Mar 13, 2014 at 5:57 PM, Toshio Kuratomi a.badger@gmail.com wrote:
On Thu, Mar 13, 2014 at 07:03:19PM -0300, "Germán A. Racca" wrote:
On 03/13/2014 06:57 PM, Toshio Kuratomi wrote:
I got busy but I started taking a look. It seems like the build scripts for that package (it's setup.py) are overriding some of the standard distutils commands with ones from astropy. I need to look at those and see if they're what's causing this issue.
-Toshio
Thanks for your time. If you prefer, you can contact me via IRC in Fedora channels, my nick is skytux.
Okay, it looks like it is a bug in astropy.
I'm attaching a patch that I applied to astropy that got the build working. I don't have a deep understanding of setuptools and distutils internals, so I'm not sure if the patch is correct or just is a good starting point for getting the code fixed.
-Toshio
On 03/14/2014 05:45 AM, Toshio Kuratomi wrote:
Ugh. Sorry -- the patch is essentially correct but I was too quick to think it was finished. The patch is hardcoding the build directory to "build/lib.linux-x86_64-2.7/". This is wrong. To finish off the patch and make it suitable for applying to astropy it needs to figure out that value from the build environment similar to what's being done in AstropyBuildPy.finalize_options. On different arches and systems it would be build/lib.linux-x86_64-2.7, build/lib.linux-x86-2.7, etc.
-Toshio
On Thu, Mar 13, 2014 at 5:57 PM, Toshio Kuratomi a.badger@gmail.com wrote:
On Thu, Mar 13, 2014 at 07:03:19PM -0300, "Germán A. Racca" wrote:
On 03/13/2014 06:57 PM, Toshio Kuratomi wrote:
I got busy but I started taking a look. It seems like the build scripts for that package (it's setup.py) are overriding some of the standard distutils commands with ones from astropy. I need to look at those and see if they're what's causing this issue.
-Toshio
Thanks for your time. If you prefer, you can contact me via IRC in Fedora channels, my nick is skytux.
Okay, it looks like it is a bug in astropy.
I'm attaching a patch that I applied to astropy that got the build working. I don't have a deep understanding of setuptools and distutils internals, so I'm not sure if the patch is correct or just is a good starting point for getting the code fixed.
-Toshio
No problem! In the mean time, I contacted the maintainer of Astropy and informed about this situation. I'm waiting for a response.
Germán.
On 03/14/2014 05:45 AM, Toshio Kuratomi wrote:
Ugh. Sorry -- the patch is essentially correct but I was too quick to think it was finished. The patch is hardcoding the build directory to "build/lib.linux-x86_64-2.7/". This is wrong. To finish off the patch and make it suitable for applying to astropy it needs to figure out that value from the build environment similar to what's being done in AstropyBuildPy.finalize_options. On different arches and systems it would be build/lib.linux-x86_64-2.7, build/lib.linux-x86-2.7, etc.
-Toshio
Issue reported upstram:
https://github.com/astropy/astropy/issues/2223
Germán.
On Thu, Mar 13, 2014 at 5:57 PM, Toshio Kuratomi a.badger@gmail.com wrote:
On Thu, Mar 13, 2014 at 07:03:19PM -0300, "Germán A. Racca" wrote:
On 03/13/2014 06:57 PM, Toshio Kuratomi wrote:
I got busy but I started taking a look. It seems like the build scripts for that package (it's setup.py) are overriding some of the standard distutils commands with ones from astropy. I need to look at those and see if they're what's causing this issue.
-Toshio
Thanks for your time. If you prefer, you can contact me via IRC in Fedora channels, my nick is skytux.
Okay, it looks like it is a bug in astropy.
I'm attaching a patch that I applied to astropy that got the build working. I don't have a deep understanding of setuptools and distutils internals, so I'm not sure if the patch is correct or just is a good starting point for getting the code fixed.
-Toshio
-- packaging mailing list packaging@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/packaging
packaging@lists.fedoraproject.org