Sorry for coming to this discussion late... I still recovering from 5 days w/out power...
Orion Poplawski wrote:
Jesse Keating wrote:
Here is another example where maintainers need to coordinate more and pay attention.
https://admin.fedoraproject.org/updates/F9/FEDORA-2008-10000 (rpcbind-0.1.7-1.fc9) went into Fedora 9 updates testing on 11-22. On the 25th bodhi feedback showed that this requires selinux changes.
https://admin.fedoraproject.org/updates/F9/FEDORA-2008-11122 (selinux-policy-3.3.1-115.fc9) went into Fedora 9 updates testing on 12-10. This was the build to fix rpcbind.
On 12-10 rpcbind was submitted for stable. On the same day, bodhi feedback requested that this update not go out until the matching selinux-policy went out to stable.
On 12-11 rpcbind went into stable.
Yeah, this is what prompted my earlier post. This also makes me wonder why I bother running updates-testing on some of my machines. Seems like half the time I report an issue, the update gets pushed to stable anyways.
The reason I moved rpcbind to stable was I saw the selinux-policy-3.3.1-115 had been built and pushed to testing... So I guess I just assumed selinux-policy would be push to stable as soon as I pushed rpcbind.... as soon as I realized that was not the case, I started to lobbing the selinux-policy maintainer to get the package pushed...
Another caveat was the people needing the new rpcbind package don't need or use selinux so I was getting pressure make the release.. After a number of days of waiting, seeing the fixed selinux package was available plus it truly was an selinux bug (all rpcbind was wanted to so as a setuid() to a non-root user) I decide to make the release... Pissing some people off and pleasing others... (the story of my life! ;-) )
Is there really a clean way of handling something like this? It was not apparent to me...
steved.