Differences between revisions 9 and 11 (spanning 2 versions)
Revision 9 as of 2009-04-23 07:52:47
Size: 1688
Editor: was
Comment:
Revision 11 as of 2009-05-07 14:21:11
Size: 4532
Comment: add Solaris link, link back to the current Sage discussion thread, timetable and gcc 4.4.0 status
Deletions are marked like this. Additions are marked like this.
Line 2: Line 2:

See https://groups.google.com/group/sage-devel/browse_thread/thread/8cadd969aca13e64 for a discussion thread about the release.
Line 16: Line 18:
== Doctest to 75% == == Timetable ==

{{{
Wed May 6: * mhansen (by 2am) -- patches/todo list so anybody can help.
           * robertwb will have reviewed david roe's
           * mabshoff provides bill with symmetrica bug testcase
           * mabshoff fix libtool issue for pynac

Thu May 7: sage-4.0.alpha0.tar -- ecl; mop up positive review; freebsd
                                    burcin updated pynac
                                    (this will better test ecl)
           * wstein, robertwb, nick, mhansen --
                        help with pynac/symbolics doctest

Fri May 8: sage-4.0.alpha1.tar -- mhansen's symbolics
                                                      get in david roe's patch

Sat May 9: Sage-4.0 day -- IRC /merge/referee:
             goal: coverage to 75.0%.
             E.g.,
               * wstein or cwitty -- referee implicit plot 3d

Sun May 10: sage-4.0.rc0.tar (feature freeze)
               * fix singular prompt issue (solaris fix)
               * fix other singular remaining issues.

Mon May 11: fix fallout; malb fix os x libsingular (?)

Tue May 12: sage-4.0.rc1.tar

Wed May 13:

Thu May 14: sage-4.0.final.tar

Fri May 15: Release sage-4.0.tar.
}}}

=== Solaris ===

See [[solaris/sage-3.4.2|Sage 3.4.2 build issues and doctest failures]].

=== gcc 4.4.0 ===

{{{
I spend some time tonight on the gcc 4.4.0 porting problem and there
is only little work left to be done: I had resolved all issues in Sage
3.1.2, but neglected to merge all the fixed into subsequent releases.

Issues from 3.1.2 (3.4.2):

 * gmp (gone in MPIR)
 * ntl (Fixed in ntl-5.4.2.p7.spkg, need to add spkg-check before
posting spkg)
 * fplll (update to 3.0.11 - or maybe a future 3.0.12 due to bug that
already existed in 2.x.)
 * polybori - fixes:
    * src/boost_1_34_1.cropped/boost/mpl/aux_/full_lambda.hpp +230
    * src/boost_1_34_1.cropped/boost/mpl/apply.hpp +138
    * src/boost_1_34_1.cropped/boost/mpl/bind.hpp +364
    * src/boost_1_34_1.cropped/boost/mpl/bind.hpp +531
    * src/boost_1_34_1.cropped/boost/mpl/apply_wrap.hpp +81
    * src/boost_1_34_1.cropped/boost/mpl/apply_wrap.hpp +173
 * clisp - *boom* - might be build problem - will use ecl in
4.0.alpha0 anyway

The main issue I ran into was a bug in libfplll that would crash the
doctest with some probability. The bug is also in the previous fplll
release and I have send a proposed fix upstream, so I am confident we
will have gcc 4.4 support in Sage 4.0. At least two people have either
complained in IRC, so the demand seems to clearly exist. :)

Cheers,

Michael
}}}

=== Doctest to 75% ===
Line 31: Line 109:
  * fix coverage script to not test nested functions: 50 doctests   * fix coverage script to not test nested functions: 50 doctests  (this is http://trac.sagemath.org/sage_trac/ticket/877)

Sage-4.0

See https://groups.google.com/group/sage-devel/browse_thread/thread/8cadd969aca13e64 for a discussion thread about the release.

In order of priority:

  1. Pynac -- mhansen+burcin will have a patch soon
  2. Solaris 10 -- about 7 remaining bugs for 32 bit Sparc and x86, some rather difficult. [to satisfy this goal, user will have to build custom toolchain, and we're only really going for this working on two specific machines on skynet]
  3. Doctest 75% -- plan given below
  4. OS X 64-bit -- doesn't start up due to libSingular issue, but after that, probably 5 tests [to satisfy this goal, it is only necessary that it is easy to build *binaries*, and it only has to support Intel. It's not necessary that any random user can just type "make" and have everything trivially work].
  5. Remove deprecated quaddouble.spkg

Good idea to fix:

  1. http://trac.sagemath.org/sage_trac/ticket/3984, and update to upstream version.

  2. Update sage-vmware-4.0.zip so that it includes Ubuntu-8.04.LTS

Timetable

Wed May 6: * mhansen (by 2am) -- patches/todo list so anybody can help.
           * robertwb will have reviewed david roe's
           * mabshoff provides bill with symmetrica bug testcase
           * mabshoff fix libtool issue for pynac

Thu May 7:   sage-4.0.alpha0.tar -- ecl; mop up positive review; freebsd
                                    burcin updated pynac
                                    (this will better test ecl)
           * wstein, robertwb, nick, mhansen --
                        help with pynac/symbolics doctest

Fri May 8:   sage-4.0.alpha1.tar -- mhansen's symbolics
                                                      get in david roe's patch

Sat May 9:   Sage-4.0 day -- IRC /merge/referee:
             goal: coverage to 75.0%.
             E.g.,
               * wstein or cwitty -- referee implicit plot 3d

Sun May 10:  sage-4.0.rc0.tar  (feature freeze)
               * fix singular prompt issue (solaris fix)
               * fix other singular remaining issues.

Mon May 11:  fix fallout;        malb fix os x libsingular (?)

Tue May 12:  sage-4.0.rc1.tar

Wed May 13:

Thu May 14:  sage-4.0.final.tar

Fri May 15:  Release sage-4.0.tar. 

Solaris

See Sage 3.4.2 build issues and doctest failures.

gcc 4.4.0

I spend some time tonight on the gcc 4.4.0 porting problem and there
is only little work left to be done: I had resolved all issues in Sage
3.1.2, but neglected to merge all the fixed into subsequent releases.

Issues from 3.1.2 (3.4.2):

 * gmp (gone in MPIR)
 * ntl (Fixed in ntl-5.4.2.p7.spkg, need to add spkg-check before
posting spkg)
 * fplll (update to 3.0.11 - or maybe a future 3.0.12 due to bug that
already existed in 2.x.)
 * polybori - fixes:
    * src/boost_1_34_1.cropped/boost/mpl/aux_/full_lambda.hpp +230
    * src/boost_1_34_1.cropped/boost/mpl/apply.hpp +138
    * src/boost_1_34_1.cropped/boost/mpl/bind.hpp +364
    * src/boost_1_34_1.cropped/boost/mpl/bind.hpp +531
    * src/boost_1_34_1.cropped/boost/mpl/apply_wrap.hpp +81
    * src/boost_1_34_1.cropped/boost/mpl/apply_wrap.hpp +173
 * clisp - *boom* - might be build problem - will use ecl in
4.0.alpha0 anyway

The main issue I ran into was a bug in libfplll that would crash the
doctest with some probability. The bug is also in the previous fplll
release and I have send a proposed fix upstream, so I am confident we
will have gcc 4.4 support in Sage 4.0. At least two people have either
complained in IRC, so the demand seems to clearly exist. :)

Cheers,

Michael 

Doctest to 75%

  • cdef -- we don't count now, but we should count cpdef
  • 70%
  • Jon Hanke -- 100 doctests
  • Mike Hansen -- 200 doctests
  • David Roe -- around 500 (!!) doctests (or more: current coverage of padics to be submitted: 91.7%)
    • --> needs review

  • modsym/ambient.py: 71 doctests
  • fix coverage script to not test nested functions: 50 doctests (this is http://trac.sagemath.org/sage_trac/ticket/877)

    • - sage_doc.py coverage gets confused.
  • algebra_*: 80 doctests
  • matrix: 200
  • tachyon.py: 53 (in plot, not interfaces)
  • remove 4ti2, cocoa, reduce, template: 70
  • doctest mathematica, matlab, kash: 80
  • delete catalogue
  • sage -coverage *
    • No functions in init.py

    • No functions in all.py
    • No functions in symmetrica.pyx

plan/sage-4.0 (last edited 2009-05-13 02:34:34 by MichaelAbshoff)