Re: Original sources, or not
On Thu, Sep 18, 2003 at 08:01:46AM +0200, Matthias Urlichs wrote:
> many packages seem to contain .orig.tar.gz files which may or may not be
> directly related to the files actually available from upstream. That is
> I think that it would make sense to add a requirement to Policy that
> the .orig.tar.gz file should be an unmodified copy from upstream.
> If that is not possible or doesn't make sense for some reason, the way to go
> from there to here should be documented either textually or, preferably,
> achieved by running debian/package_upstream (or whatever).
This is already mentionned in Policy 4.8:
This target fetches the most recent version of the original
source package from a canonical archive site (via FTP or WWW, for
example), does any necessary rearrangement to turn it into the
original source tar file format described below, and leaves it in
the current directory.
I have no problem with documenting *why* the source is not pristine,
but I refuse to be forced to document *how* to get the non pristine
one from the pristine one, unless you accept procedure like:
2)Look at each file whether it is DFSG free or not. If not, delete it.
This cannot be automated.
Most developers are lazy and will write a script to automate the process
if they can. In this case debian/get-orig-source is the policy
documented way to go.
Developers use the changelog to document change to the upstream tarball
and I feel it is quite proper.
I think we should rather make a best practice document on why and how
repackage upstream tarball. IMHO, this should be done more often, to
save bandwidth and disk space.
Reasons I know :
--- tarball does not exists, or is in a stupid format.
--- Files have stupid permissions.
--- tarball contains files at the root.
--- Some files are not DFSG free.
--- You can't add binary files (e.g. icons) in a diff. Using uuencode
is not optimal. Sometimes it is better to sneak them in the source
--- tarball include large stuff that we don't want to package.
--- tarball contains 90% arch:all code and 10% arch:any code.
In this case it is better to split the tarball in two source
package, one arch:all and one arch:any, so you dont need to
unpack 20Mb of junk to build a 50kB program.
--- Some part of the tarball are pretty static. Again it is better
to split the tarball than to upload again and again the same static
--- People use DBS and put the tarball in a larger tarball. I object on
this practice, but it is irrelevant to this discussion.
In conlusion, I don't see any need to change policy.