[Debichem-devel] Jmol again

[Michael Banck]

Hi,

first off, sorry for the late approval of your post, I've added you to
the debichem-devel whitelist now.

Also, CCing Egon as he was involved in prior discussion of jmol
packaging.

On Mon, Aug 16, 2010 at 10:05:03PM +0200, Georges Khaznadar wrote:
> I would like to package Jmol, because it would be useful as a
> recommended package for two other packages which I maintain: wims and
> chemical-structures.

I see wims on packages.debian.org, but not chemical-structures, where is
that one maintained?
 
> I downloaded the source from the SVN repository (branch 12.0) today, and
> I could check how easy it was to build a Debian package, technically
> speaking. 

The current Debian packaging is here:

svn://svn.debian.org/svn/debichem/unstable/jmol/debian

I just updated the debian/copyright file with what I had.

I can add you to the alioth debichem team if you tell me your alioth ID,
or request to join.

> However there is a big license issue, as the sources contain a
> handful of jars coming from third parties, and only two of them seem to
> exist in already available packages.
 
Which ones?

So far I succeeded (I think) in building jmol by doing a 

find . -name "*.jar" | xargs rm -f

in the clean target, followed by

ant -Dlib.dir=/usr/share/java -Dvecmath.jar=vecmath.jar
-Ditext.jar=itext1.jar -Dcommons-cli.jar=commons-cli.jar

in the build target, using libcommons-cli-java, libjaxp1.3-java,
libsaxon-java, libvecmath-java, libitext1-java as Build-Depends, see the
above packaging.

However, there are also a couple of code-drops in the src/ directory
which are 3rd party, see below.

The other problem is that jmol's build system seems to combine all those
external jars into the main libjmol-java instead of linking to them, and
(as I am no Java expert) I have not found a work-around for this yet.

Anyway, my memory is a bit fuzzy on that as it has been a couple of
weeks.

> I saw that the packaging of Jmol had been discussed in the past in this
> mailing list. Had some solutions been suggested to package dependencies,
> or were some bad licenses discovered in them?

I have been going through them, and so far, I did not find any obviousy
non-free licenses.  However, Debconf of sort blocked further movement
here.

Personally, I don't think the non-linking of the jars is a very big
issue, however it would be nice to fix it.  Unfortunately, squeeze has
been frozen since, so we got lots of time to do so...


cheers,

Michael