[Debichem-devel] Pymol patches and suggestions

[Michael Banck]

On Sun, Feb 01, 2009 at 01:19:13AM +0100, Daniel Leidert wrote:
> - 04_Rules.make.dpatch
> 
> For what is it used? Is it still used or is Rules.make obsolete by
> setup.py? The python versions in this file look very old and a build
> without this patch succeeds.

OK - could well be bitrot here.  ISTR Rules.make was needed (and had to
be manually edited) in order to build the C code.  That was probably
before there was configure.ac and this is now obsolete.
 
> - 05_examples_data_path.dpatch
> 
> Uses "$PYMOL_DATA_PATH" which should be "$PYMOL_DATA/demo", right? 

Yeah, I guess; this is probably a leftover from some reshuffling.

> Is the `import os' necessary here for the absolute path? Sorry if this
> is a dumb question. I don't know much about Python.
 
Sorry, I don't remember why I/whoever else put it there.  As those are
with the demos and they are easy to test (just fire up pymol and click
through them and see whether they all work) I suggest removing those
imports for a test and see whether things still work.

> - 11_fix__cmd_import.dpatch
> 
> Which problem did this one solve? It looks like a patch, which could be
> suggested to upstream.

This one was introduced in the 0.98+0.99rc6-1.1 NMU so I don't know the
rationale either; I assume it posed problems at the time.
 
> - 13_activate_vmd_plugin.dpatch
> 
> Can we maybe provide a switch-based solution to upstream to get rid of
> this patch? I'm not familiar with setup.py writing, but maybe someone
> else is(?).

Sounds sensible, maybe put it in the packages' TODO?
 
> - 09_chempy_data_path.dpatch, 17_cmyk_png_data_path.dpatch
> 
> I wonder, if we should change the installation strategy for
> data/{chempy,pymol} and do not install these directories
> as /usr/share/{chempy,pymol}. Instead we could install these directories
> as /usr/share/pymol/data/{chempy,pymol}. The rationale behind is,
> that /usr/share/pymol/date IMO is similar to $PYMOL_DATA even in
> upstreams code. We could then suggest to upstream to use consequently
> $PYMOL_DATA instead of $PYMOL_PATH/data in its source. So we can get rid
> of the above patches and IMHO this will make the upstream code a bit
> cleaner for distribution packagers and upstream, because then one can
> consequently rely on $PYMOL_PATH and $PYMOL_DATA (if both are different
> like in our case) without needing to patch files.

OK - this FHS and pymol-data shuffling around was a huge trial 'n error
thing.  I'm all for unifying this with upstream (and we already had
some success about this in the past)

> Forwarded to upstream (bug tracker):
> - 06_remove_shebang.dpatch
> - 14_script_not_executable.dpatch
> - 16_pymem_del.dpatch
> 
> I further created an item in the upstream bug-tracker to remove
> executable permissions from data files. So we can remove the chmod calls
> in debian/rules.
 
Thanks a lot for this!


Michael