[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: For those who care about their packages in Ubuntu



Matt Zimmerman <mdz@ubuntu.com> writes:

> I believe there are still packages which break when bin-NMU'd (e.g.,
> Depends: = ${Source-Version}), and there are parts of our infrastructure
> which do not support them (Ubuntu doesn't do bin-NMUs).  

That's correct.  These are bugs, and should be reported and fixed.

Since you don't do bin-NMU's, you could simply alter the version of
every package to add an "ubuntu" tag, and then be done with it, right?
That would work well and be very easy to implement.

> Why is it now important to you that the version numbers be changed, though?
> This is only an issue when mixing packages between different derivatives,
> which already breaks in other subtle ways, so I'm not very much inclined to
> try to un-break it in this particular way, given that it's non-trivial.

What is non-trivial?  Adding the version tag?  

You're populating the net with .debs which *look* just like the
official Debian ones, but *aren't*, and the result is that confusion
can happen.  I believe that someone has already posted about how this
confusion has come up in at least one case.

In addition, there are many times people have reported that the Debian
practice of changing version numbers for successive binNMU's has been
of great assistance in tracking down bugs.

>> Will you establish an Ubuntu policy that all bugs found, whether
>> patched or not, which might exist in the upstread Debian package,
>> should always be reported to the BTS?
>
> The "might" here is a problem.  It is considered to be in poor taste to
> report bugs to bugs.debian.org which have not been verified on Debian, and
> attempting to confirm every bug is not feasible for us (we can't even come
> close to confirming every bug on Ubuntu).

How about whenever you actually fix a bug then?  Or how about using
one's best judgment?

In any case, I want to note what has just happened here.  You received
a clear, easily implemented, request about what would be a wonderful
contribution, and which is (from the Debian perspective) entirely
non-controversial.

Your response was not to say "yes, great!" but rather to argue about
whether these are things Debian should care about, and protest that
they are not valuable and take too much work.

Thomas



Reply to: