Bugzilla Bug 213616

As the native install of pymol provides a plugin interface for apbs, it should
depend. Second for the right functionality of apbs with pymol two env variables
should be set. From the ideas discussed in bug #213346 I rewrote the ebuild a
little bit and introduced the USE flag apbs. It makes pymol depend on
sci-chemistry/apbs and for the USE="-apbs" it removes the plugin.

Additionally I added the python_mod_cleanup in the pkg_postrm section.

Reproducible: Always

Created an attachment (id=146328) [details]
pymol-1.0-r2.ebuild

Created an attachment (id=146934) [details]
pymol-1.0-r2.ebuild

Made it some more portable.

Hi Justin,

That looks good, thanks much for your work. In general, it is better 
to post a diff to an exisiting ebuild rather than the ebuild itself since
it is much easier to spot the differences that way! Unless of course
the changes are very significant.

I think you can pull the line
APBS_VER="$(best_version sci-chemistry/apbs | cut -d/ -f2)"
into the "if use apbs" conditional later on.

Also, I will try to have apbs install its files into /usr/share/apbs
rather than apbs-<version>. Otherwise pymol will hard depend on a
particular apbs version forcing a recompile of pymol 
every time apbs will be bumped. In that case you can get rid
of the APBS_VER alltogether and replace it by just apbs.
I'll post back once this is done.

cheers,
Markus

(In reply to comment #3)
> I think you can pull the line
> APBS_VER="$(best_version sci-chemistry/apbs | cut -d/ -f2)"
> into the "if use apbs" conditional later on.
It must be in in the conditional, because there is no reason to have a
"best_version sci-chemistry/apbs". I changed the lines and at a diff.

Created an attachment (id=146996) [details]
pymol-1.0-r1.ebuild patch

Now a diff.

Cool, thanks!

I changed apbs to install into /usr/share/apbs. 
I did not bump abps to a new revision though, hence users that 
have apbs already installed will need to recompile it to pick up
this change. You may want to add a few einfo lines pointing
this out (probably wrapped in a "if use apbs") to the ebuild.

Thanks,
Markus

Hi Markus!

I changed the lines and added the ewarn. Hope the test case is legal. If not,
please correct me.

Thanks,
Justin

Created an attachment (id=147156) [details]
pymol.diff

Created an attachment (id=147706) [details]
pymol.diff

Forgott some dependencies.

Fixed in the science overlay. I noticed your ebuild had some apbs variables
commented out, so could you clarify the status of bug #213346 on that bug?

The updated version of the apbs plugin allows to define default locations
inside the plugin. I set the right paths there so that there is no need to set
them somewhere else. BUG #213346 is obsolete now.