Re: Science group on alioth
On Thu, 2008-01-24 at 22:02 +0100, Christophe Prud'homme wrote:
> [ Thursday 24 January 2008 ]
> | Le jeudi 24 janvier 2008 à 20:31 +0100, Christophe Prud'homme a écrit :
> | > [ Thursday 24 January 2008 ]
> | >
> | > | Hello,
> | > |
> | > | After a few discussions with Raphael and Adam, I would like to propose
> | > | to create a pkg-science on alioth in order to create a dynamic around
> | > | scientific software packaging.
> | > | Adam created a package of Opencascade. To simplify the maintenance of
> | > | this package and to create of future packages like Code_Aster or
> | > | Salome, we could create a "task force" on Alioth.
> | >
> | > there is pkg-scicomp on Alioth for scientific computing software
> | > netgen and gmsh which have interface with opencascade are managed there
> | Hello Christophe,
> | In fact, I thought this was a dead group when I had a look to:
> | http://lists.alioth.debian.org/pipermail/pkg-scicomp-commits/
> | But after a quick look, I found it is not at all the case
> | (pkg-scicomp-devel). Sorry for not looking deeper.
> | Can we import Opencascade inside and other projects ? (if you don't mind
> | the increase of traffic).
> | I really don't mind giving up the idea of pkg-science (Raphael already
> | created the project and the svn... Sorry for this).
> I don't mind at all and I actually would be very happy about this
> I was meaning to work on opencascade and salomé for some time,
> since I am about to use them on an ANR project
> The policy is not to include the sources, just the diff. Would that be ok ?
> We use svn-buildpackage but that is not mandatory
That sounds fine. My .diffs are all in the debian directory so the
patches are cleanly separated and labeled.
It would be great to have some team maintenance for these packages.
The OpenCASCADE package is 99% complete and builds fine on unstable. It
is missing two .dat files required for Salomé, and I have been waiting
for some free time to add in Jason's scripts to split up the binary
packages in a more sane way (maybe tomorrow?). It's also *almost*
lintian-clean! So I think that is nearly ready to upload, the only
question is: main or non-free... For a discussion of this, see:
Salomé is also 99% complete (my -4 is building now, and should be ready
to post in a few hours), but is very lintian-unclean, and has dependency
* It needs two extra netgen symbols, and needs to load the netgen
libraries at runtime, so those need to be shared; I sent patches
via bugs 461929 and 462105 to do these. Haven't heard back from
* I like OpenMPI, so I patched HDF5 to use it, and sent the patch
via bug 457080, which the maintainers accepted. But the
resulting package is taking a while because of an interface
* Salomé needs omniORB 4.0.x which is no longer in unstable (but
is in testing); porting it to 4.1.x looks like it will take a
very long time. I tried to contact upstream via their web form,
but haven't heard back.
So although Salomé is LGPL, it will be a while before can include it in
Debian. Also, because it links with OCC (OCC license), and Qt (GPL),
the OCC license will need to be deemed GPL-compatible in order for
Salomé to be distributable. :-(
GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6
Engineering consulting with open source tools