Upstream bugs vs. Fedora bugs: KDE people do it wrong

[Michael Schwendt]

On Mon, 29 Mar 2010 23:25:50 -0400, Orcan wrote:

> > What is RH bugzilla for, when not using for bugs in fedora?
> 
> For packaging related bugs, or bugs related to Fedora specific
> customizations on packages.

What you call "Fedora specific customizations on packages" may be also the
chosen combination of software releases, which may be specific to Fedora.
For example, as in "upstream developers have tested with different library
releases and a software environment that differs from the Fedora distribution".
It wouldn't be the first time someone asks whether some piece of software
in Fedora is patched or not.

"RH bugzilla" is Fedora's distribution-specific issue tracking system
where to inform the distribution's package maintainers about defects of
their compiled product. Fedora package maintainers should have an interest
in learning about all problems a Fedora user runs into with the software
packaged and distributed by Fedora. Not every issue may be important
enough to justify a package update/upgrade, but you do want to receive
feedback about whether the software, you offer in form of packages, does
not work. In the same way, you would also have an interest in learning
whether upstream's next release would work "better" and be an improvement.

> Sometimes a user files a bug that I can't even reproduce. It is better
> for the user to report such bugs upstream than to me. When I forward
> the bug upstream, upstream asks me to test certain things and to
> report back. It doesn't make sense to test something for a bug that I
> can't even reproduce. Why should I play the middle man? It's
> inefficient.

It's similary inefficient if you expect upstream to support the bug
reporter with Fedora-specific help. While you would be able to teach 
the reporter how to install missing debuginfo packages or point him
at Fedora Wiki pages, the upstream developer may not be able to do that
and might be less willing to spend time on it.

It becomes worse in case upstream cannot reproduce a problem either.
That bears a higher risk that upstream will start to think it's
just something Fedora-specific that's broken.

> Nevertheless, I try to do my best to spare time to locate the bug in
> many cases.  But I don't think this should be my duty.

The term "duty" is poorly chosen when referring to volunteers. The size
of the Fedora package collection would be cut down to a third, if packagers
were obliged to do "certain things" which have not been mandatory before.
Still there are some expectations towards volunteer package maintainers.

> It takes time
> to go through someone else's code and figure out what is wrong. I
> don't always have this time.

Of course not. There are other scenarios, too, such as a bug-fix that
would require you to either rewrite/redesign several files or try to
backport big modifications of the source code (with or without interface
changes), if you could not upgrade to upstream's development tree.