Entering freeze for libvirt-6.8.0
I have just tagged v6.8.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/ Please give the release candidate some testing and in case you find a serious issue which should have a fix in the upcoming release, feel free to reply to this thread to make sure the issue is more visible. If you have not done so yet, please update NEWS.rst to document any significant change you made since the last release. Thanks, Jirka
On Thu, 2020-09-24 at 15:13 +0200, Jiri Denemark wrote:
I have just tagged v6.8.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/
Can we please stop generating and publishing the source RPMs? They're of little use considering that you can easily run $ rpmbuild -ta libvirt-x.y.z.tar.xz to obtain RPMs from a release archive, so all they really do is clutter the directory listing.
If you have not done so yet, please update NEWS.rst to document any significant change you made since the last release.
Yes, please do that everybody :) -- Andrea Bolognani / Red Hat / Virtualization
On Thu, Sep 24, 2020 at 18:38:18 +0200, Andrea Bolognani wrote:
On Thu, 2020-09-24 at 15:13 +0200, Jiri Denemark wrote:
I have just tagged v6.8.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/
Can we please stop generating and publishing the source RPMs? They're of little use considering that you can easily run
$ rpmbuild -ta libvirt-x.y.z.tar.xz
to obtain RPMs from a release archive, so all they really do is clutter the directory listing.
The nice thing about the source RPMs is that they are internally signed (in contrast to an external signature in *.asc for tarballs). Personally I don't find them useful either, but some people may think otherwise and I'm not sure they will raise their voice here. We could just stop creating source RPMs and wait if anyone complains. Or we could be a little bit more conservative and just move them to a subdirectory perhaps. I don't mind either way. Jirka
On Thu, Sep 24, 2020 at 3:00 PM Jiri Denemark <jdenemar@redhat.com> wrote:
On Thu, Sep 24, 2020 at 18:38:18 +0200, Andrea Bolognani wrote:
On Thu, 2020-09-24 at 15:13 +0200, Jiri Denemark wrote:
I have just tagged v6.8.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/
Can we please stop generating and publishing the source RPMs? They're of little use considering that you can easily run
$ rpmbuild -ta libvirt-x.y.z.tar.xz
to obtain RPMs from a release archive, so all they really do is clutter the directory listing.
The nice thing about the source RPMs is that they are internally signed (in contrast to an external signature in *.asc for tarballs). Personally I don't find them useful either, but some people may think otherwise and I'm not sure they will raise their voice here. We could just stop creating source RPMs and wait if anyone complains. Or we could be a little bit more conservative and just move them to a subdirectory perhaps. I don't mind either way.
I like that Source RPMs are produced, because I can trivially use them in my import->build->test pipeline, since my build system accepts SRPMs as input and not random tarballs. :) So at least here's one voice for keeping them. :) -- 真実はいつも一つ！/ Always, there's only one truth!
On Thu, Sep 24, 2020 at 03:39:03PM -0400, Neal Gompa wrote:
On Thu, Sep 24, 2020 at 3:00 PM Jiri Denemark <jdenemar@redhat.com> wrote:
On Thu, Sep 24, 2020 at 18:38:18 +0200, Andrea Bolognani wrote:
On Thu, 2020-09-24 at 15:13 +0200, Jiri Denemark wrote:
I have just tagged v6.8.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/
Can we please stop generating and publishing the source RPMs? They're of little use considering that you can easily run
$ rpmbuild -ta libvirt-x.y.z.tar.xz
to obtain RPMs from a release archive, so all they really do is clutter the directory listing.
The nice thing about the source RPMs is that they are internally signed (in contrast to an external signature in *.asc for tarballs). Personally I don't find them useful either, but some people may think otherwise and I'm not sure they will raise their voice here. We could just stop creating source RPMs and wait if anyone complains. Or we could be a little bit more conservative and just move them to a subdirectory perhaps. I don't mind either way.
I like that Source RPMs are produced, because I can trivially use them in my import->build->test pipeline, since my build system accepts SRPMs as input and not random tarballs. :)
If you have the tarball you can trivially generate the source RPM if you need it because the tarball still contains the .spec file. IOW just run $ rpmbuild -ts libvirt-6.7.0.tar.xz Wrote: /home/berrange/rpmbuild/SRPMS/libvirt-6.7.0-1.fc32.src.rpm Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
On Fri, Sep 25, 2020 at 4:35 AM Daniel P. Berrangé <berrange@redhat.com> wrote:
On Thu, Sep 24, 2020 at 03:39:03PM -0400, Neal Gompa wrote:
On Thu, Sep 24, 2020 at 3:00 PM Jiri Denemark <jdenemar@redhat.com> wrote:
On Thu, Sep 24, 2020 at 18:38:18 +0200, Andrea Bolognani wrote:
On Thu, 2020-09-24 at 15:13 +0200, Jiri Denemark wrote:
I have just tagged v6.8.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/
Can we please stop generating and publishing the source RPMs? They're of little use considering that you can easily run
$ rpmbuild -ta libvirt-x.y.z.tar.xz
to obtain RPMs from a release archive, so all they really do is clutter the directory listing.
The nice thing about the source RPMs is that they are internally signed (in contrast to an external signature in *.asc for tarballs). Personally I don't find them useful either, but some people may think otherwise and I'm not sure they will raise their voice here. We could just stop creating source RPMs and wait if anyone complains. Or we could be a little bit more conservative and just move them to a subdirectory perhaps. I don't mind either way.
I like that Source RPMs are produced, because I can trivially use them in my import->build->test pipeline, since my build system accepts SRPMs as input and not random tarballs. :)
If you have the tarball you can trivially generate the source RPM if you need it because the tarball still contains the .spec file. IOW just run
$ rpmbuild -ts libvirt-6.7.0.tar.xz Wrote: /home/berrange/rpmbuild/SRPMS/libvirt-6.7.0-1.fc32.src.rpm
That assumes I have the ability to add such a mechanism that can do that automatically. :) -- 真実はいつも一つ！/ Always, there's only one truth!
On Fri, 2020-09-25 at 05:09 -0400, Neal Gompa wrote:
On Fri, Sep 25, 2020 at 4:35 AM Daniel P. Berrangé <berrange@redhat.com> wrote:
On Thu, Sep 24, 2020 at 03:39:03PM -0400, Neal Gompa wrote:
I like that Source RPMs are produced, because I can trivially use them in my import->build->test pipeline, since my build system accepts SRPMs as input and not random tarballs. :)
If you have the tarball you can trivially generate the source RPM if you need it because the tarball still contains the .spec file. IOW just run
$ rpmbuild -ts libvirt-6.7.0.tar.xz Wrote: /home/berrange/rpmbuild/SRPMS/libvirt-6.7.0-1.fc32.src.rpm
That assumes I have the ability to add such a mechanism that can do that automatically. :)
A testing pipeline that can only work with SRPMs seems quite limiting, and will not be able to deal with many subprojects such as for example Go and Rust language bindings... We actually only produce SRPMs for the main C library IIRC. Anyway, while I personally think that we should not be generating SRPMs, I can't deny they seem to offer some value to you, and so I will begrudgingly retire my motion to stop publishing them :) -- Andrea Bolognani / Red Hat / Virtualization
On Fri, Sep 25, 2020 at 4:35 AM Daniel P. Berrangé <berrange@redhat.com> wrote:
If you have the tarball you can trivially generate the source RPM if you need it because the tarball still contains the .spec file. IOW just run
$ rpmbuild -ts libvirt-6.7.0.tar.xz Wrote: /home/berrange/rpmbuild/SRPMS/libvirt-6.7.0-1.fc32.src.rpm
I've never tried this before... what does it do if the .spec file is non-trivial and includes things expanded by autoconf, or if there are multiple different .spec files? Personally, I have found the libvirt provided spec file limiting - and use a distro one like Fedora as a starting point instead. And, I meant to try and figure out why this was and provide feedback about it. But, for other projects such as openvswitch - I find that the one they build and maintain upstream, to be very useful. In this case, you do run autoconf and configure, and "make rpm-fedora" generates both RPM and SRPM. From that respect, the need to distribute a .src.rpm separately is not as important, but the ability to build an SRPM using an easy mechanism is the real requirement and I'm mostly wondering if "rpmbuild -ts" is sophisticated enough to support many real-world use cases, or if it is only useful for the most trivial and eventually limiting use cases. -- Mark Mielke <mark.mielke@gmail.com>
On 9/25/20 2:30 PM, Mark Mielke wrote:
On Fri, Sep 25, 2020 at 4:35 AM Daniel P. Berrangé <berrange@redhat.com <mailto:berrange@redhat.com>> wrote:
If you have the tarball you can trivially generate the source RPM if you need it because the tarball still contains the .spec file. IOW just run
$ rpmbuild -ts libvirt-6.7.0.tar.xz Wrote: /home/berrange/rpmbuild/SRPMS/libvirt-6.7.0-1.fc32.src.rpm
I've never tried this before... what does it do if the .spec file is non-trivial and includes things expanded by autoconf, or if there are multiple different .spec files?
That's not relevant in the case of libvirt, because we no longer use autoconf/configure/make, but use meson instead, and the source tarfile only has a single .spec file. Even when libvirt did use autoconf, that step was done on the file libvirt.spec.in as a part of make dist, and so the source tar has always contained a libvirt.spec that is ready to be used directly by rpmbuild.
Personally, I have found the libvirt provided spec file limiting - and use a distro one like Fedora as a starting point instead.
I'm not sure what difference you're seeing here. The libvirt.spec in Fedora git is just directly imported from the upstream libvirt source tar every time the package is rebased. There is no difference. (well, except that the Fedora libvirt.spec file maintains the incomplete and thus totally pointless %changelog section).
And, I meant to try and figure out why this was and provide feedback about it.
If there's something you think is missing in libvirt's spec file, definitely send a patch for it.
But, for other projects such as openvswitch - I find that the one they build and maintain upstream, to be very useful. In this case, you do run autoconf and configure, and "make rpm-fedora" generates both RPM and SRPM. From that respect, the need to distribute a .src.rpm separately is not as important, but the ability to build an SRPM using an easy mechanism is the real requirement and I'm mostly wondering if "rpmbuild -ts" is sophisticated enough to support many real-world use cases, or if it is only useful for the most trivial and eventually limiting use cases.
Hmm. Maybe you're asking the question "in general" for any arbitrary package, while I am (and Dan was) answering only relative to libvirt? In that case I should stop and let someone who knows what they're talking about take over :-)
On Fri, Sep 25, 2020 at 02:30:32PM -0400, Mark Mielke wrote:
On Fri, Sep 25, 2020 at 4:35 AM Daniel P. Berrangé <berrange@redhat.com> wrote:
If you have the tarball you can trivially generate the source RPM if you need it because the tarball still contains the .spec file. IOW just run
$ rpmbuild -ts libvirt-6.7.0.tar.xz Wrote: /home/berrange/rpmbuild/SRPMS/libvirt-6.7.0-1.fc32.src.rpm
I've never tried this before... what does it do if the .spec file is non-trivial and includes things expanded by autoconf, or if there are multiple different .spec files?
It only works when there is a single .spec file, which is the case with libvirt tarballs. Expansion of macros is largely irrelevant until you get the build stage.
Personally, I have found the libvirt provided spec file limiting - and use a distro one like Fedora as a starting point instead.
Errm, the libvirt.spec inside the tarball *IS* the Fedora distro spec. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
I have just tagged v6.8.0-rc2 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/ Please give the release candidate some testing and in case you find a serious issue which should have a fix in the upcoming release, feel free to reply to this thread to make sure the issue is more visible. If you have not done so yet, please update NEWS.rst to document any significant change you made since the last release. Thanks, Jirka
participants (6)
-
Andrea Bolognani -
Daniel P. Berrangé -
Jiri Denemark -
Laine Stump -
Mark Mielke -
Neal Gompa