ace optional spkg

[William Stein]

Hi,

I would like to remove the ACE optional spkg from the list of optional spkg at sagemath.org, because:

 (1) I just tried to build it on ubuntu 12.04 and it failed

 (2) The SPKG.txt file is empty (see http://sagemath.org/packages/optional/) so I don't even know what it does or who the maintainer is!

Is anybody opposed?  

William

--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org

[leif]

William Stein wrote:
> Hi,
>
> I would like to remove the ACE optional spkg from the list of optional

> spkg at sagemath.org <http://sagemath.org>, because:

>
> (1) I just tried to build it on ubuntu 12.04 and it failed
>
> (2) The SPKG.txt file is empty (see
> http://sagemath.org/packages/optional/) so I don't even know what it
> does or who the maintainer is!

There's a three years old "needs info" ticket on ACE:

http://trac.sagemath.org/sage_trac/ticket/8551

(It does have an "updated" spkg, too... :-) )


-leif

> Is anybody opposed?
>
> William

--

() The ASCII Ribbon Campaign
/\ Help Cure HTML E-Mail

[Dima Pasechnik]

On Friday, 26 April 2013 02:18:22 UTC+8, leif wrote:

William Stein wrote:
> Hi,
>
> I would like to remove the ACE optional spkg from the list of optional
> spkg at sagemath.org <http://sagemath.org>, because:
>
>   (1) I just tried to build it on ubuntu 12.04 and it failed
>
>   (2) The SPKG.txt file is empty (see
> http://sagemath.org/packages/optional/) so I don't even know what it
> does or who the maintainer is!

It is a fast implementation of coset enumeration algorithms (for f.p. groups), with a GAP interface.

  

There's a three years old "needs info" ticket on ACE:

http://trac.sagemath.org/sage_trac/ticket/8551  

 

(It does have an "updated" spkg, too... :-) )

it's broken just the same, but it should be trivial to fix.

It works as a GAP package just fine, and is a "standard" GAP package.

  

-leif


> Is anybody opposed?
>
> William

I'm for fixing it, rather than removing.

Dima

[William Stein]

On Thu, Apr 25, 2013 at 6:44 PM, Dima Pasechnik <dim...@gmail.com> wrote:

> Is anybody opposed?
>
> William

I'm for fixing it, rather than removing.

Dima

What about removing it as an optional package right now... since it doesn't work, then adding it back later once that ticket is closed?

I can't see positive value in hosting a broken optional package with no information about it. 

William

[Jeroen Demeyer]

On 04/26/2013 03:55 AM, William Stein wrote:

> What about removing it as an optional package right now... since it
> doesn't work, then adding it back later once that ticket is closed?
>
> I can't see positive value in hosting a broken optional package with no
> information about it.

I agree with this. Harald: could you remove that package for now?

[Dima Pasechnik]

sure, this way is fine - I misinterpreted "remove" somehow.

There are more non-working optional packages, in fact.

E.g. 4ti2 does not install (at least not on OSX 10.6).

Dima

 

William

[Harald Schilly]

On Fri, Apr 26, 2013 at 9:00 AM, Jeroen Demeyer <jdem...@cage.ugent.be> wrote:
> I agree with this. Harald: could you remove that package for now?

done, it's in the "archive" where old ones spend the remainder of
their lives ;-)

H

[Volker Braun]

On Friday, April 26, 2013 9:51:45 AM UTC+1, Dima Pasechnik wrote:

E.g. 4ti2 does not install (at least not on OSX 10.6).

I think it doesn't compile with any modern g++

[leif]

Where "modern" means GCC >= 4.7.0 (including 4.8.0, and presumably at
least all recent versions of Clang). Its 'configure' is also broken
with these in that it doesn't "find" GMP/MPIR.

See http://trac.sagemath.org/sage_trac/ticket/13498


Note that also Sage 5.9's doctesting framework breaks its doctests.


-leif

[William Stein]

I propose splitting the optional packages into two groups: tested and untested.  Anybody interested in helping?

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscribe@googlegroups.com.
To post to this group, send email to sage-...@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.

[leif]

William Stein wrote:
> I propose splitting the optional packages into two groups: tested and
> untested. Anybody interested in helping?

Well, "tested" is a moving (and parameterized) target.

It at least seems we should more regularly test them (ideally upon
"releasing" a new Sage release candidate, and when new major compiler
versions get out).

The release managers would certainly be happy to do that as well
(although we could automate that for the "major" platforms), but IMHO
the so-called spkg maintainers are in charge here, too. (There are
spkgs which are meanwhile effectively unmaintained, though; these should
probably get moved to "untested" or "unmaintained", until someone
volunteers to further maintain them.)


-leif

> On Apr 26, 2013 6:07 AM, "leif" <not.r...@online.de

> <mailto:not.r...@online.de>> wrote:
>
> Volker Braun wrote:
>
> On Friday, April 26, 2013 9:51:45 AM UTC+1, Dima Pasechnik wrote:
>
> E.g. 4ti2 does not install (at least not on OSX 10.6).
>
>
> I think it doesn't compile with any modern g++
>
>
> Where "modern" means GCC >= 4.7.0 (including 4.8.0, and presumably
> at least all recent versions of Clang). Its 'configure' is also
> broken with these in that it doesn't "find" GMP/MPIR.
>

> See http://trac.sagemath.org/sage___trac/ticket/13498

[William Stein]

On Fri, Apr 26, 2013 at 7:37 AM, leif <not.r...@online.de> wrote:

William Stein wrote:

I propose splitting the optional packages into two groups: tested and
untested.  Anybody interested in helping?

Well, "tested" is a moving (and parameterized) target.

Testing is something we define.  Once it is defined it is not moving and is not parametrized until we choose to change the definition.    Defining "tested" for our purposes is part of the project.  It will be very clearly defined at the top of the list of such packages.  It can start with: "each of the following packages *built* on 64-bit Ubuntu 12.04 with sage-5.8 and the following Ubuntu packages installed"... which would be *massively* better than the current situation, where many of the optional  packages don't work (?).   After that is done, we could move on to asserting that not only do the packages install on that platform, but a specific test suite passed (usually "sage -t --optional=package"). 

Also, it would be good to distinguish optional packages that users shouldn't install unless they have a special need (e.g., valgrind), since they maybe rebuild Python sub-optimally.  

  

It at least seems we should more regularly test them (ideally upon "releasing" a new Sage release candidate, and when new major compiler versions get out).

The release managers would certainly be happy to do that as well (although we could automate that for the "major" platforms), but IMHO the so-called spkg maintainers are in charge here, too.  (There are spkgs which are meanwhile effectively unmaintained, though; these should probably get moved to "untested" or "unmaintained", until someone volunteers to further maintain them.)

Yes, they would all get moved out.

I'm interested in doing this, because I install Sage on a lot of machines, including the new https://cloud.sagemath.org stuff I'm rolling out soon, and want to install "all sensible optional spkg"s for users.   I should be able to do this by typing one line, but instead, I have to carefully look through each spkg, try to figure out if it is: (1) some old crap meant to deal with the lack of packages on OS X (say), (2) math related, (3) broken, etc.   Instead, I should be able to safely issue one command, wait a bit, and get all recommended and tested optional packages. 

 -- William 

 

-leif

On Apr 26, 2013 6:07 AM, "leif" <not.r...@online.de

<mailto:not.r...@online.de>> wrote:

    Volker Braun wrote:

        On Friday, April 26, 2013 9:51:45 AM UTC+1, Dima Pasechnik wrote:

             E.g. 4ti2 does not install (at least not on OSX 10.6).


        I think it doesn't compile with any modern g++


    Where "modern" means GCC >= 4.7.0 (including 4.8.0, and presumably
    at least all recent versions of Clang).  Its 'configure' is also
    broken with these in that it doesn't "find" GMP/MPIR.

    See http://trac.sagemath.org/sage___trac/ticket/13498

    <http://trac.sagemath.org/sage_trac/ticket/13498>


    Note that also Sage 5.9's doctesting framework breaks its doctests.


    -leif

--
() The ASCII Ribbon Campaign
/\   Help Cure HTML E-Mail

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscribe@googlegroups.com.
To post to this group, send email to sage-...@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.

[leif]

William Stein wrote:
> I propose splitting the optional packages into two groups: tested and
> untested. Anybody interested in helping?

We should probably start by updating

http://wiki.sagemath.org/Sage_Spkg_Tracking

which also lists optional spkgs (but currently only five of them!),
collecting maintainers (and upstream contacts), then asking them /
checking whether the information is still current.

[Jeroen Demeyer]

On 04/26/2013 03:40 PM, William Stein wrote:
> I propose splitting the optional packages into two groups: tested and
> untested.

What's the difference between untested optional and experimental? If
there is none, then move them to experimental.

[Dima Pasechnik]

4ti2 version 1.5 quietly lives in http://www.math.ucdavis.edu/~latte/software/latte-integrale-1.5.3.tar.gz

on http://www.math.ucdavis.edu/~latte/software.php 

That version 1.5 still does not work on OSX, but at least it works on Linux, as far as I recall.

Dima

[kcrisman]

I think this is preferable to archiving, which is probably more for projects that have been subsumed within something else, as opposed to ones that we just don't have a maintainer for etc.