Re: perl packaging replication
On Mon, Oct 25, 2004 at 08:48:25PM +0100, Stephen Quinney wrote:
>I noticed today several new packages have appeared in sid which seem to
>exactly or closely replicate those already available in the perl
>package itself. I am wondering what the point of this is and whether
>it is a good idea or not? In the past bugs have been submitted for the
>removal of perl module packages that are replicating work done in the
>main perl package.
>
>package name | perl | package
>------------------|------------|--------
>libfile-spec-perl | 0.87 | 0.90
>libfile-temp-perl | 0.14 | 0.14
>libtext-wrap-perl | 2001.09291 | 2001.0929
>
>Of these 3 only one of them seems justified and, to me, it would seem
>better to request an update of the libfile-spec-perl within the perl
>package rather than introduce another new package.
The right way to update modules in core is to create a new package
rather than updating the perl package--Perl Policy specifically allows,
and indeed encourages this (see §4.1).
The recent 5.8.4-2.3 NMU of perl for MIME::Base64 would have been better
achieved by uploading an updated libmime-base64-perl for example.
Packaging core modules in this way does introduce maintenance issues:
since dpkg doesn't support versioned depends, you will need to specify a
dependency initially on "libfoo-perl (>= X)", and subsequently change
that dependency to "perl (>= Y)" when a perl version is released which
contains version X or newer of the Foo module.
This is obviously an issue only when a specific version is required.
Other packages which depend on "libfoo-perl" without a version are fine.
Raphael:
* There is no reason to have the .orig.tar.gz contain a copy of the
upstream tarball rather than just *being* the upstream tarball.
While this may be one interpretation of Debian policy's requirement
that it "contain the source code from the upstream authors of the
program", I don't believe (abominations such as DBS notwithstanding)
that was the intention.
* Your debian/rules is rather byzantine. Is there a reason why the
rules suggested in Perl policy are inadequate?
--bod
Reply to: