[csw-maintainers] Implications of publishing packages to unstable
rupert at opencsw.org
Sun Sep 18 13:20:57 CEST 2011
would it be possible to include an experimental stage to the automated
release process? i.e. have a machine where these packages are
automatically deployed and can be run at least once before promoting
them to unstable.
2011/9/18 Maciej (Matchek) Bliziński <maciej at opencsw.org>:
> Dear maintainers,
> I'd like to clarify the implications of publishing packages to unstable.
> When you call the csw-upload-pkg tool, certain things happen, and it's
> important that you understand what these things are.
> 1. The .pkg files are uploaded to allpkgs
> 2. checkpkg is run against the packages; if the check fails, the procedure stops
> 3. The packages are added to the unstable catalog
> http://mirror.opencsw.org/opencsw/unstable/ - unstable is an official,
> signed catalog which gets pushed to mirrors all around the world, and
> gets installed by users
> 4. Packages from unstable are installed on the buildfarm during
> periodic manual sync
> csw-upload-pkg tool should not be taken lightly. You must not push
> packages that you haven't at least smoke-tested. If you push broken
> packages, you cause grief on the buildfarm. You can also cause grief
> by pushing backward-incompatible packages that are part of the
> infrastructure, e.g. any of the CSWgar-dev package dependencies.
> If you have built packages that you would like to just try out,
> unstable is not the right place for that purpose. Use the experimental
> catalogs instead, by copying files into /home/experimental/$LOGNAME
> and visiting http://buildfarm.opencsw.org/experimental.html
> maintainers mailing list
> maintainers at lists.opencsw.org
> .:: This mailing list's archive is public. ::.
More information about the maintainers