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

Re: Package naming rules.



>>>>> "B" == Bill Mitchell <mitchell@mdd.comm.mot.com> writes:

B> Personally, I think that we should stick with the guideline of
B> using the upstream author's unmangled package name.  I'm not a big
B> fan of tailoring requirements to make implementation more
B> convenient.  Also, as Ian J. has recently remarked, renaming
B> packages is a pain.  I don't think that we should lightly make a
B> change which would require renaming 77 packages which are currently
B> in distribution.

I'll change my packages to whatever we agree on, but I think that
*consistency* and ease of parsing filenames is a really good idea.
Note that being too tied to the upstream maintainer's name is often
not desirable.  Consider my perl-tk package.  The upstream name was
"Tk" which is obviously not reasonable for Debian.

Personally I prefer using some simple naming convention that is
totally unambiguous.  I think the recent proposal to replace dashes in
the package name with underscores when creating the package files
solves the problem (even if it does have the disadvantage of making
the package name in the file name slightly different from the package
name in the package).

For those who missed it the first time around, this would mean that a
package named some-package would have files named:

some_package-version-revision.arch.deb
some_package-revision.tar.gz
some_package-revision.diff.gz

I think If we use this scheme, we should probably add the converse
restriction that package names (not the resultant file names) cannot
contain underscores.  Otherwise it would be difficult to be able to
automatically map (uniquely) from a file name to the corresponding
package name.

I also think that having packages whose package names cannot be
readily predicted from the file name (like ldso, shellutils, and
syslogd) is a really bad idea.  Do we gain anything for the added
complexity and confusion?  I can just see the repeated questions to
debian-user and the resulting FAQ entry now.  What's the point?

In any case, whatever we're going to do, let's do it soon.  I predict
we are going to have a huge influx of new users soon, and after that
kind of widening in distribution, changes like this will be much more
agonizing.

--
Rob


Reply to: