Bug#252101: ITP: gromacs -- versatile package to perform molecular dynamics
On Sat, Nov 05, 2005 at 12:42:04PM +0100, Michael Banck wrote:
> I took a look at your gromacs package as I accidently found the ITP.
> I've been trying to package gromacs for a couple of years now, so I am
> glad I checked before just uploading something :)
>
> Anyway, I played around with your package a bit and got some questions:
>
> 1. Did you consider putting the changes you do to upstream into seperate
> patches, maintained in a debian/patches directory via a patch-system
> like dpatch or quilt?
Didn't seem necessary. Nearly the entire set of patches relates to
documentation changes or additions that I'm pushing upstream, which should
shrink the non-debian-specific components of the diff to nearly nothing
if they're accepted.
> 2. When you modify configure, why do you not change configure.in
> accordingly? I believe we should rather patch configure.in and
> regenerate configure (possibly removing unneeded noise in the diff).
> Did you consider sending your modifications upstream, possible making
> them more generic (like ${MPI_LIBSUFFIX})
I'm sure I had some reason for not editing configure.in when I did it,
but I cannot for the life of me remember what that reason was. You're
right that it should hold the patch rather than configure.
The reason for the Debian-specific name is to make it blindingly obvious
that it's a change from upstream. Because scientific software demands
reproducibility, the authors prefer it if all changes from the upstream
source are clearly marked; I doubt that particular change will affect
anyone, but it doesn't hurt to highlight it. I will be sending that
patch upstream as well with a more generic name.
> 3. Did you consider making a gromacs-doc package including the manual?
> I think it bloats up the Debian diff rather considerably. Alos, you got
> the manual from www.gromacs.org, right? Did you clarify whether the
> copyright is the same as for the program or something else?
The .pdf is the big killer in the .diff at the moment; as mentioned,
all the other doc changes are heading upstream. The authors of the .pdf
manual have informed me that it's licensed under the GPL, identically
with the rest of the package.
Splitting off -doc packages is always something of a judgment call
between space efficiency and Packages.gz bloat. The total of
/usr/share/doc/gromacs and /usr/share/gromacs/tutor (the tutorial files)
is only 4.3MB, out of a total installed size of ~16MB. Since this isn't
that much of a savings, I decided it wasn't worth splitting off a
separate -doc package.
> 4. Finally, would you consider maintaining gromacs as part of a
> hopefully-soon-to-be-started Debian-chemistry-packaging-team effort from
> svn.debian.org? Are you a Debian Developer? If not, do you need
> sponsoring?
I am certainly amenable to team maintenance, and am greatly interested in
assisting with other chemistry software. I'm not a DD, and do need a
sponsor for the package. I've asked Andreas Tille (cc:'d on this message)
already, but he sounds fairly busy at the moment -- perhaps the two of you
could discuss that?
--
Nicholas Breen
nbreen@ofb.net
Reply to: