Debian support for vdsm and subpackages
We currently manage debian packaging files under https://gerrit.ovirt.org/#/q/project:releng-tools instead of in the project itself. imo it makes it harder, as developer won't update the debian folder while changing spec parts , which makes much more observation work for Simone.. But if we keep to manage debian that way we should remove the debian folder from cpopen, vdsm, pthreading to avoid duplication. Simone can you explain more how you manage to flow code changes and update the debian support currently? should I indeed drop the current debian code that we have in vdsm? -- *Yaniv Bronhaim.*
On Sun, Dec 13, 2015 at 2:25 PM, Yaniv Bronheim <ybronhei@redhat.com> wrote:
We currently manage debian packaging files under https://gerrit.ovirt.org/#/q/project:releng-tools instead of in the project itself. imo it makes it harder, as developer won't update the debian folder while changing spec parts , which makes much more observation work for Simone.. But if we keep to manage debian that way we should remove the debian folder from cpopen, vdsm, pthreading to avoid duplication.
Simone can you explain more how you manage to flow code changes and update the debian support currently? should I indeed drop the current debian code that we have in vdsm?
-- *Yaniv Bronhaim.*
Issue with keeping debian subdir within the package itself is that it's totally useless unless maintained. When a new release is issued just tagging the code won't be enough anymore because debian changelog expect to be updated with the release version. So tagging vdsm-4.17.14 without having 4.17.14 in the changelog will cause the package to fail the build (while rpm just issue a error in rpmlint... ) having a separate repo decouple the distribution packaging from the package itself and make it easier to handle at release time. -- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
Issue with keeping debian subdir within the package itself is that it's totally useless unless maintained. When a new release is issued just tagging the code won't be enough anymore because debian changelog expect to be updated with the release version. So tagging vdsm-4.17.14 without having 4.17.14 in the changelog will cause the package to fail the build (while rpm just issue a error in rpmlint... )
having a separate repo decouple the distribution packaging from the package itself and make it easier to handle at release time.
While I have very little stake in this, I have to say that looking at this from the side, it looks strange that RPM is supported as a primary build target for all projects while DEB needs a separate repo. Maybe the build scripts should be made smarter and work around Debians limitations (For example let the makefile dynamically generate the changelog form the tags)? -- Barak Korren bkorren@redhat.com RHEV-CI Team
On Mon, Dec 14, 2015 at 12:26 PM, Barak Korren <bkorren@redhat.com> wrote:
Issue with keeping debian subdir within the package itself is that it's totally useless unless maintained. When a new release is issued just tagging the code won't be enough
because debian changelog expect to be updated with the release version. So tagging vdsm-4.17.14 without having 4.17.14 in the changelog will cause
package to fail the build (while rpm just issue a error in rpmlint... )
having a separate repo decouple the distribution packaging from the
anymore the package
itself and make it easier to handle at release time.
While I have very little stake in this, I have to say that looking at this from the side, it looks strange that RPM is supported as a primary build target for all projects while DEB needs a separate repo. Maybe the build scripts should be made smarter and work around Debians limitations (For example let the makefile dynamically generate the changelog form the tags)?
rpm build works just because rpm is not as strict as debian. there's a reason if fedora, rhel, centos uses dist-git with external spec file instead of rely on the spec file we ship within the tarball. The spec included in the tarball is not production / enterprise level ready.
-- Barak Korren bkorren@redhat.com RHEV-CI Team
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
On Mon, Dec 14, 2015 at 1:59 PM, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
On Mon, Dec 14, 2015 at 12:26 PM, Barak Korren <bkorren@redhat.com> wrote:
Issue with keeping debian subdir within the package itself is that it's totally useless unless maintained. When a new release is issued just tagging the code won't be enough anymore because debian changelog expect to be updated with the release version. So tagging vdsm-4.17.14 without having 4.17.14 in the changelog will cause the package to fail the build (while rpm just issue a error in rpmlint... )
having a separate repo decouple the distribution packaging from the package itself and make it easier to handle at release time.
While I have very little stake in this, I have to say that looking at this from the side, it looks strange that RPM is supported as a primary build target for all projects while DEB needs a separate repo. Maybe the build scripts should be made smarter and work around Debians limitations (For example let the makefile dynamically generate the changelog form the tags)?
rpm build works just because rpm is not as strict as debian. there's a reason if fedora, rhel, centos uses dist-git with external spec file instead of rely on the spec file we ship within the tarball. The spec included in the tarball is not production / enterprise level ready.
Strange, I understood from Eyal that we push the spec vdsm generates to dist-git, and this is the spec used for the build. Patches to make the spec production/ enterprise ready are welcome :-)
-- Barak Korren bkorren@redhat.com RHEV-CI Team
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
participants (4)
-
Barak Korren -
Nir Soffer -
Sandro Bonazzola -
Yaniv Bronheim