future of blas/lapack/atlas packages?

[Matthias Klose-2]

Hi all,

blas and lapack are basic numeric libraries much used within Debian;
the same holds for atlas which seems to be be needed by more and more
packages.  We had some difficulties getting these packages converted
to build with gfortran; initial packages were made for experimental,
then some packages were provided by the package maintainer, which were
further updated to the current state in unstable/testing.  However it
looks like these packages are basically unmaintained, and the
packaging is an "interesting" one, in that many people did fail to
upgrade the atlas package from 3.6 to 3.8. How could we go on with the
packaging of these packages?  Camm currently doesn't seem to have
enough time, so is there anybody interested in maintainance who would
be able to update the atlas package and maybe set up a packaging group
which could be open for the current package maintainer?

thanks, Matthias


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Ondrej Certik]

Hi,

thanks Matthias for raising this issue. I am interested in helping with
this.

We need to maintain the packages in a svn/git repository and I suggest
we simply create new packges, something like

libatlas3.8-*

or something. We get it compile, we get it to debian and we start fixing
all potential problems. Let's have both atlas 3.6 and atlas 3.8 and
users will simply use those, that work. When we get 3.8 in a good shape,
we may then for example remove the libatlas-* packages and provide them
with libatlas3.8-* etc.

Ondrej


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Matthias Klose-2]

Ondrej Certik writes:

> Hi,
>
> thanks Matthias for raising this issue. I am interested in helping with
> this.
>
> We need to maintain the packages in a svn/git repository and I suggest
> we simply create new packges, something like
>
> libatlas3.8-*
>
> or something. We get it compile, we get it to debian and we start fixing
> all potential problems. Let's have both atlas 3.6 and atlas 3.8 and
> users will simply use those, that work. When we get 3.8 in a good shape,
> we may then for example remove the libatlas-* packages and provide them
> with libatlas3.8-* etc.

please don't rename the package without reason. uploads to
experimental should do fine here.

  Matthias


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Ondrej Certik]

On Tue, Jul 8, 2008 at 4:06 PM, Matthias Klose <doko@...> wrote:

> Ondrej Certik writes:
>> Hi,
>>
>> thanks Matthias for raising this issue. I am interested in helping with
>> this.
>>
>> We need to maintain the packages in a svn/git repository and I suggest
>> we simply create new packges, something like
>>
>> libatlas3.8-*
>>
>> or something. We get it compile, we get it to debian and we start fixing
>> all potential problems. Let's have both atlas 3.6 and atlas 3.8 and
>> users will simply use those, that work. When we get 3.8 in a good shape,
>> we may then for example remove the libatlas-* packages and provide them
>> with libatlas3.8-* etc.
>
> please don't rename the package without reason. uploads to
> experimental should do fine here.

Yes, I forgot about experimental. Yes, that would work perfectly fine.

Ondrej


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Dirk Eddelbuettel]

(Doku and Camm re-added to CC --Dirk)

On 8 July 2008 at 16:02, Ondrej Certik wrote:
| Hi,
|
| thanks Matthias for raising this issue.

Seconded!  

This has always been a point of great pride for Debian, and I have provided
Atlas to users of R and Octave (when I still maintained the latter) for
almost eight years.

Camm did great and pioneering work here.  We should build on this.
 
| I am interested in helping with this.

I may be out of my depth with the compiler / porting issues and am thinly
stretched with > 100 packages (including ess which 'we' [as in a new small
group] took from Camm recently).

| We need to maintain the packages in a svn/git repository

Sounds good. I will try to help, at least with testing.  Anybody else?

Dirk

--
Three out of two people have difficulties with fractions.


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Ondrej Certik]

Hi Dirk!

glad to see you around.

On Wed, Jul 9, 2008 at 3:07 AM, Dirk Eddelbuettel <edd@...> wrote:

>
> (Doku and Camm re-added to CC --Dirk)
>
> On 8 July 2008 at 16:02, Ondrej Certik wrote:
> | Hi,
> |
> | thanks Matthias for raising this issue.
>
> Seconded!
>
> This has always been a point of great pride for Debian, and I have provided
> Atlas to users of R and Octave (when I still maintained the latter) for
> almost eight years.

Exactly. I am a python guy, so we need this in python-numpy and
python-scipy packages.
Recently, the Sage project, that also includes R, uses atlas3.8, but
we don't have it in Debian.

>
> Camm did great and pioneering work here.  We should build on this.
>
> | I am interested in helping with this.
>
> I may be out of my depth with the compiler / porting issues and am thinly
> stretched with > 100 packages (including ess which 'we' [as in a new small
> group] took from Camm recently).
>
> | We need to maintain the packages in a svn/git repository
>
> Sounds good. I will try to help, at least with testing.  Anybody else?

Unfortunately, I am also very busy these days, so I am not promising
anything. But if I do any work, I'll post to these lists, so that you
can test it.

Do you have experience with git packaging? I don't, but maybe we could
try it, as it could be easier for all of us -- no need to joing any
team, just git collab main.

Ondrej


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[David Bremner-4]

>>>>> "Ondrej" == Ondrej Certik <ondrej@...> writes:

    Ondrej> Do you have experience with git packaging? I don't, but
    Ondrej> maybe we could try it, as it could be easier for all of us
    Ondrej> -- no need to joing any team, just git collab main.

Debian science is also using git, and you are welcome there I am sure.
I'm not sure how much team activity is going on so far, but it will be
no worse than collab maint, and maybe better in the long run.  On the
other hand, it is hard to think of something more appropriate to
pkg-scicomp than Atlas.

David





--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Ondrej Certik]

On Wed, Jul 9, 2008 at 4:50 PM, David Bremner <bremner@...> wrote:

>>>>>> "Ondrej" == Ondrej Certik <ondrej@...> writes:
>
>    Ondrej> Do you have experience with git packaging? I don't, but
>    Ondrej> maybe we could try it, as it could be easier for all of us
>    Ondrej> -- no need to joing any team, just git collab main.
>
> Debian science is also using git, and you are welcome there I am sure.
> I'm not sure how much team activity is going on so far, but it will be
> no worse than collab maint, and maybe better in the long run.  On the
> other hand, it is hard to think of something more appropriate to
> pkg-scicomp than Atlas.

Ok, thanks for all the emails. As I said I am too busy now to start
packaging it, but if I'll do any work, I'll let you know.

Ondrej


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...

[Michael Banck]

On Wed, Jul 09, 2008 at 03:27:35PM +0200, Ondrej Certik wrote:
> > | We need to maintain the packages in a svn/git repository
> >
> > Sounds good. I will try to help, at least with testing.  Anybody else?
>
> Do you have experience with git packaging? I don't, but maybe we could
> try it, as it could be easier for all of us -- no need to joing any
> team, just git collab main.

Why not put it into the pkg-scicomp svn repository?  It fits perfectly
there, AFAICT.


Michael


--
To UNSUBSCRIBE, email to debian-science-request@...
with a subject of "unsubscribe". Trouble? Contact listmaster@...