Sage 3.0.5 now available in Debian sid

[Tim Abbott]

Hello all,

As of today, Sage 3.0.5 is available in Debian sid, so you can now run
"apt-get install sagemath" and get Sage installed on your Debian
system[1].

However, the sage package has bitrotted substantially during the 7 months
it was hanging out in the NEW queue, so I'd not count on it to work. In
fact, Ondrej Certik has already reported it segfaulting at startup on
i386. Bugs are being tracked (as usual) at
<http://bugs.debian.org/sagemath> and we're currently using
sage-...@googlegroups.com for related discussion. I plan send mail
here again once I upload a version that is believed working on x86.

I'm planning to begin work on upgrading to 3.3 this weekend, since I
believe that will help with some of the problems.

It looks like there are a couple of likely complications:

(1) Sage 3.3 alphas seem to be using an svn revision of MPIR. Is it
possible to use GMP instead? Or alternatively, is there a stable MPIR
release yet?

(2) Sage seems to contain an spkg for "pynac", which is apparently a
python-integrated gynac, where Sage is the upstream source? Does it have
anything to do with this pynac: <http://sourceforge.net/projects/pynac/>?

-Tim Abbott

[1] Well, if your system is running one of the architectures on which Sage
compiles. Apparently, Sage builds on s390 but not MIPS; who'd have known?
The apparent problem with MIPS is the use of long double functions like
sqrtl in the Sage library (in sage/combinat/partitions_c.cc, and maybe
nowhere else). MIPS support is quite low on my priority list, but if
you're interested, feel free to send me patches.

[mabshoff]

On Feb 1, 4:23 pm, Tim Abbott <tabb...@MIT.EDU> wrote:
> Hello all,

Hi Tim,

> As of today, Sage 3.0.5 is available in Debian sid, so you can now run
> "apt-get install sagemath" and get Sage installed on your Debian
> system[1].

Excellent.

> However, the sage package has bitrotted substantially during the 7 months
> it was hanging out in the NEW queue, so I'd not count on it to work.  In
> fact, Ondrej Certik has already reported it segfaulting at startup on
> i386.  Bugs are being tracked (as usual) at
> <http://bugs.debian.org/sagemath> and we're currently using
> sage-...@googlegroups.com for related discussion.  I plan send mail
> here again once I upload a version that is believed working on x86.

Ok.

> I'm planning to begin work on upgrading to 3.3 this weekend, since I
> believe that will help with some of the problems.
>
> It looks like there are a couple of likely complications:
>
> (1) Sage 3.3 alphas seem to be using an svn revision of MPIR.  Is it
> possible to use GMP instead?  Or alternatively, is there a stable MPIR
> release yet?

It is possible to use GMP, but you will see doctest failures with
mathematically equivalent, but correct results. A stable MPIR release
should be out very soon, right now the only bugs fixed are the ones
that pop up when using --enable-fat on 64 bit Intel and AMD CPUs
(which never worked in GMP).

> (2) Sage seems to contain an spkg for "pynac", which is apparently a
> python-integrated gynac, where Sage is the upstream source?  Does it have
> anything to do with this pynac: <http://sourceforge.net/projects/pynac/>?

No, that is different. The upstream code is in the spkg. It is also
about to be updated to 1.1.2 IIRC.

>         -Tim Abbott
>
> [1] Well, if your system is running one of the architectures on which Sage
> compiles.  Apparently, Sage builds on s390 but not MIPS; who'd have known?
> The apparent problem with MIPS is the use of long double functions like
> sqrtl in the Sage library (in sage/combinat/partitions_c.cc, and maybe
> nowhere else).  MIPS support is quite low on my priority list, but if
> you're interested, feel free to send me patches.

Strange, when I build Sage 3.1.2 or so on a MIPS64 SiCortex box I do
not recall running into that problem. But it is no problem to use MPFR
there and so that problem will go away.

Cheers,

Michael

[Burcin Erocal]

Hi,

On Sun, 1 Feb 2009 16:41:30 -0800 (PST)
mabshoff <mabs...@googlemail.com> wrote:

> On Feb 1, 4:23 pm, Tim Abbott <tabb...@MIT.EDU> wrote:

> > (2) Sage seems to contain an spkg for "pynac", which is apparently a
> > python-integrated gynac, where Sage is the upstream source?  Does
> > it have anything to do with this pynac:
> > <http://sourceforge.net/projects/pynac/>?
>
> No, that is different. The upstream code is in the spkg. It is also
> about to be updated to 1.1.2 IIRC.

The updated package is linked to from #5096.

I made a couple of changes to the automess scripts so pynac can be
installed with ginac without the two overlapping. Since this is not a
real problem in the Sage distribution, I can't say that we've tested
this though. Let me know if there is any problem and I'll try to fix it
properly.

Cheers,

Burcin

[William Stein]

On Mon, Feb 2, 2009 at 12:31 AM, Burcin Erocal <bur...@erocal.org> wrote:
>
> Hi,
>
> On Sun, 1 Feb 2009 16:41:30 -0800 (PST)
> mabshoff <mabs...@googlemail.com> wrote:
>
>> On Feb 1, 4:23 pm, Tim Abbott <tabb...@MIT.EDU> wrote:
>
>> > (2) Sage seems to contain an spkg for "pynac", which is apparently a
>> > python-integrated gynac, where Sage is the upstream source? Does
>> > it have anything to do with this pynac:
>> > <http://sourceforge.net/projects/pynac/>?
>>\
>> No, that is different. The upstream code is in the spkg. It is also
>> about to be updated to 1.1.2 IIRC.

Just to emphasize this -- that Pynac above is a project that is _dead_
-- there has been zero activity on that sourceforge page since 2001.
I made up the name "Pynac" for the Pythonized version of Ginac.

William

[Tim Abbott]

I mentioned that I'd send mail here again once the segfault problems are
fixed.

I believe that both of the crash-on-start problems with Sage 3.0.5 in
Debian are fixed in version 3.0.5dfsg-2, which is available in Debian sid.

-Tim Abbott

[Ondrej Certik]

On Sun, Feb 8, 2009 at 7:40 PM, Tim Abbott <tab...@mit.edu> wrote:
>
> I mentioned that I'd send mail here again once the segfault problems are
> fixed.

I want to congratulate you once again for all the work you did. I
noticed on your webpage, that you were told that it would need 5 or 6
people working on it. It was my estimate at SD6 in Bristol, given my
current working load, that I would need 5 more Ondrejs. So I am *very*
impressed by the work you did.

> I believe that both of the crash-on-start problems with Sage 3.0.5 in
> Debian are fixed in version 3.0.5dfsg-2, which is available in Debian sid.

Indeed, sage now starts on both i386 and amd64. Unfortunately, maxima
doesn't play well neither on i386 or amd64:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=514648

so any calculus doesn't work yet.

Ondrej

[mabshoff]

On Feb 9, 12:40 pm, Ondrej Certik <ond...@certik.cz> wrote:
> On Sun, Feb 8, 2009 at 7:40 PM, Tim Abbott <tabb...@mit.edu> wrote:

Hi Tim, Ondrej,

> > I mentioned that I'd send mail here again once the segfault problems are
> > fixed.
>
> I want to congratulate you once again for all the work you did. I
> noticed on your webpage, that you were told that it would need 5 or 6
> people working on it. It was my estimate at SD6 in Bristol, given my
> current working load, that I would need 5 more Ondrejs. So I am *very*
> impressed by the work you did.

+1 :)

> > I believe that both of the crash-on-start problems with Sage 3.0.5 in
> > Debian are fixed in version 3.0.5dfsg-2, which is available in Debian sid.
>
> Indeed, sage now starts on both i386 and amd64. Unfortunately, maxima
> doesn't play well neither on i386 or amd64:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=514648
>
> so any calculus doesn't work yet.

Yeah, those are the things I am always concerned about since anything
using pexpect just breaks too easily. When switching to a different
Maxima release and also using a different lisp things get worse (even
though they shouldn't be broken like *this*) and Debian seems to use
gcl for Maxima.

In the next couple weeks I will make an effort fix

* fix the "trivial" issues Tim reported a while back, i.e. #3686-
#3690
* fix the non-owner ought to be able to doctest issues, i.e. #5155 so
that users can run the regression suite of Sage without owning any of
the Sage tree

> Ondrej

Cheers,

Michael

[Ondrej Certik]

and:

* get rid of maxima for basic calculus

Ondrej

[mabshoff]

On Feb 9, 1:43 pm, Ondrej Certik <ond...@certik.cz> wrote:
> On Mon, Feb 9, 2009 at 1:17 PM, mabshoff <mabsh...@googlemail.com> wrote:

<SNIP>

> >  * fix the "trivial" issues Tim reported a while back, i.e. #3686-
> > #3690
> >  * fix the non-owner ought to be able to doctest issues, i.e. #5155 so
> > that users can run the regression suite of Sage without owning any of
> > the Sage tree
>
> and:
>
> * get rid of maxima for basic calculus

Well, that won't be happening in the next 10 days or so until Sage 3.4
until you are willing to share the time machine you might have been
hiding from me :)

Burcin did fix a number of issues with pynac in 3.3, so hopefully we
will see the switchover to the new code there soon. But for now
integration and limits will remain in the hands for Maxima until
someone puts some serious effort into this. And it won't be an extra
10 days to get to the level we will require here.

> Ondrej

Cheers,

Michael

[Pablo De Napoli]

Therse are really excelent news!, that will make Sage much more easy
to install on Linux systems.
Many thanks!!

Pablo