On Fri, Feb 26, 2010 at 08:41:07PM +0100, Till Maas wrote:
On Fri, Feb 26, 2010 at 07:18:58PM +0000, Matthew Garrett wrote:
> On Fri, Feb 26, 2010 at 08:15:43PM +0100, Till Maas wrote:
>
> > 1) to fix a bug or add a feature the maintainer experienced/uses
>
> If nobody is complaining about the bug, then fixing the bug can wait
> until the next Fedora release.
Why even fix the bug in Fedora at all then, if the maintainer needs to
create his own sub-distro with updated packages? Also why is the
maintainer always nobody?
The maintainer's convenience is significantly less important than that
of the end-users.
> > 2) As already told several times, not having people to test
something
> > does not mean that the package is not used
>
> If they're not complaining, they're presumably happy with the current
> state of the package?
Please come back to reality. They can also be too frustrated to report
the bug in Fedora, if it is already fixed upstream. And why should
people hit bugs again in Fedora that are already fixed upstream?
Propose a mechanism that allows you to ensure that new upstream releases
do not include any new bugs. We *know* that there are users who are
frustrated because it's difficult to follow a stable Fedora release
without things breaking. What we're discussing is a mechanism to reduce
the pain there, not one that makes it impossible for people to get their
hands on newer versions of software if the maintainer has packaged them.
> > 3) It allows new users of the package not to find/debug the
bugs again that
> > are already fixed upstream
>
> If they're willing to debug, why are they not willing to test?
Since the users are new, they are not yet there to test a package. But I
would also not be interested to test old packages just to find out that
the bugs I found are fixed in a newer release. And this already hit me
several times. I wanted to do something, installed the Fedora package,
found a bug, and realise that the bug is fixed in a new upstream
release. The only benefit of Fedora in this case is that I can easier
build the new package for me, because the spec is already there.
I'm confused now. If the maintainer hasn't uploaded a newer version,
then requiring testing for updates makes no difference. If the
maintainer has, why would the new user be testing an old version?
--
Matthew Garrett | mjg59(a)srcf.ucam.org