Tuesday, May 27, 2008

dsage.setup() broken in Sage 3.0.2

Hello folks,

unfortunately a severe DSage bug slipped by since it was neither caught by doctesting nor the DSage unit tests. We are tracking the issue at #3311 and already have a patch that fixes the issue here. While poking around two more DSage bugs [#3312 and #3314] come to light, but those two are unreviewed so far. We will accelerate the 3.0.3 release because of the above bugs, but in case you are currently using 3.0.2 and want to use DSage please apply at least the fix from #3311 I linked above. At the ticket are more fixes, but only apply the patch linked since there are dependencies between the other patches.

Cheers,

Michael

Sunday, May 25, 2008

Sage 3.0.2 released

Hi,

Sage 3.0.2 has been released. This is another bug fix release with some new features while we are waiting on the coercion rewrite to finish. Interesting new features include (in no particular order):

  • Self-orthogonal Binary Codes (Robert Miller)
  • Notebook Improvements (William Stein, Tom Boothby)
  • Portability of Sage to 64 bit OSX and Cygwin (Michael Abshoff, William Stein)
  • Posets and Semi-Lattices (Peter Jipsen and Franco Saliola)
  • Frobby for monomial ideals (Bjarke Hammersholt Roune)
As usual check the Sage 3.0.2 release tour in the wiki for details. Some people still have to fill in the details, but I send those people emails to remind them ;)

Sources were posted yesterday and today we posted 15 binaries for various platforms in the usual place. The binaries have been mirrored out to some websites and the other mirros should catch up in the next twentyfour hours. Not all binaries are there yet, i.e. the VMWare binary is still missing, but should show up by Monday at the latest.

Cheers,

Michael

Friday, May 23, 2008

Sage 3.0.2.rc3 released

Hello folks,

here we go with 3.0.2.rc3. It is basically a bunch of bug fixes for rlm's codes code that had a couple small issues left ;) In addition there is one pbuild issue fix that Jaap encountered and that in the past was also hit by David Joyner. You can download the Sage 3.0.2.rc3 sources here.

Unless something goes horribly wrong this will be identical to the final 3.0.2 release. If you added some feature that should be in the Sage release tour please add them to Sage 3.0.2 release tour wiki page. Right now that list consists of

  • Franco Saliola and Peter Jipsen's posets and semi-lattive patch
  • Robert Miller's self-orthogonal binary codes
  • Bjarke Hammersholt Roune's Frobby is now an optional spkg

But feel free to add items. As usual please test and report any issue you hit.

Cheers,

Michael

Sage 3.0.2.rc0 released

Hello folks,

this is 3.0.2.rc0, the likely final release of the 3.0.2 series and hopefully next to identical to the final 3.0.2. What is new?
  • Franco Saliola and Peter Jipsen's posets and semi-lattive patch
  • Robert Miller's self-orthogonal binary codes
  • Bjarke Hammersholt Roune's Frobby is now an optional spkg
  • Pbuild should now pass the doctests since #3097 has been fixed
In addition there were the usual bug fixes. Please build and doctest as usual and report any issues you see.

Sources and a sage.math-only binary in the usual places:


Cheers,

Michael

Merged in rc0:

#1762: Robert Miller, Michael Abshoff: Create optional graphviz package
#2121: Robert Miller: move libecm wrapper from interfaces to libs
#2519: Franco Saliola, Peter Jipsen: Add support for posets, semi-lattices, etc. to Sage
#3018: Bjarke Hammersholt Roune: Integrate Frobby into Sage
#3097: Gary Furnish, Michael Abshoff: pbuild: make sure the files from setup.py's scripts section are copied
#3104: William Stein: pbori.pyx: Make some doctest long since it uses a lot of RAM
#3112: Robert Miller: Generate self-orthogonal binary codes
#3148: Francis Clarke: improved orthogonal functions
#3218: Michael Abshoff: fix 64 bit OSX build support for mercurial
#3219: William Stein: upgrade to gmp-4.2.2 while we wait for MPIR
#3242: Robert Miller: Fix little bug in G.relabel() for G a graph
#3245: Mike Hansen: provide coefficient and coefficients methods for symbolic expressions
#3257: Gary Furnish: Pbuild ignores gcc specific default settings
#3263: Craig Citro: typo in lseries_ell.py
#3266: William Stein: Sage 3.0.2.alpha1: doctest failure in sage/server/simple/twist.py
#3267: Michael Abshoff: Sage 3.0.2.alpha1: doctest failure in sage/server/support.py
#3269: Jason Bandlow: Improve documentation for combinat/dyck_word.py
#3270: Robert Miller: trivial 100x speedup in coding theory
#3272: Craig Citro: Bug in sparse polynomials over finite fields
#3273: Robert Bradshaw: extend isqrt to work for Python int's in addition to Sage integers and objects with an isqrt method
#3274: Michael Abshoff: OSX: delete libpng*.la since we also nuke libpng*.dylib
#3275: Craig Citro: Make SL2Z distinct

Monday, May 19, 2008

Sage 3.0.2.alpha1 released

Hello folks,

the 3.0.2 release cycle so far has been a little slower than usual and there are a couple contributing factors. While some of us are waiting on the coercion rewrite to progress, others had obligations with school and the end of the semester seems to have kept a lot of people busy. 3.0.2.alpha1 contains *a lot* of porting work for Cygwin and Mac OSX 64 bit. We also merged Dan's new Weyl Character implementation and the usual set of bug fixes.

Since it has been nearly two weeks it is now time to get the release out of the door. So the plan now is to get an rc0 out the door tomorrow and then do a release by Tuesday or Wednesday. We will hopefully merge some more porting fixes, i.e. there are about ten more OSX 64 bit tickets ready to be reviewed. Other than that we have 70+ tickets with patches, so there is plenty to review. Once I catch some sleep I will do another post with details.

Sources and Binaries are in the usual place:


Right now there are two notebook related doctest failures which do not have tickets yet. Please build, test and report issues you see.

Cheers,

Michael

Details for closed tickets in alpha1:

#406: William Stein: notebook -- make tab completion not stick gap. when using the notebook in gap mode
#637: William Stein: notebook improvement -- upload allow txt worksheets.
#1733: William Stein: notebook bug -- %foo (or anything else) in a cell by itself (with nothing else in the cell) does not give an error but it *should*
#1864: William Stein: simple notebook bug -- typing ? in a comment yields introspection but shouldn't
#1892: William Stein: notebook -- uploading a data file should give some help about the DATA variable
#2359: William Stein: notebook -- make it so when you send a kill signal to the notebook server it saves state
#2636: William Stein: notebook -- changing a cell without evaluate should put the red line back to the left
#2860: William Stein: easy-to-fix bug in html.py
#2884: William Stein, Tom Boothby: notebook -- bug; @interact cell eval doesn't clear out the old html output
#2992: William Stein: notebook -- help(foo) in the notebook should not word wrap
#3024: William Stein: notebook -- parses tracebacks in the output of docstrings of help command
#3050: Timothy Clemans: notebook -- add a "remember me" checkbox to the login page
#3051: Dan Bump, Mike Hansen: Implement Weyl Characters
#3053: William Stein: notebook -- new cell_resize doesn't respect %hide at the beginning of a cell
#3069: William Stein: notebook -- typeset checkbox doesn't work after save/reload
#3137: Yi Qiang: view command in misc/latex.py -- fix to not hardcode xdvi command
#3153: Carl Witty: make finite_field_ntl_gf2e use randstate framework
#3155: Timothy Clemans: notebook postdata and behaviour of archive, delete and stop buttons
#3160: Emily Kirkman, Robert Miller: change is_planar for graphs to return bool
#3161: Michael Abshoff: sdist: #3046 seems to have broken sage-banner
#3170: Michael Abshoff: add 64 bit OSX build support to readline
#3171: Michael Abshoff: add 64 bit OSX build support to termcap
#3172: Michael Abshoff: add 64 bit OSX build support to prereq and bzip
#3176: Michael Abshoff: add 64 bit OSX build support to sqlite
#3177: Michael Abshoff: fix 64 bit OSX build support for python
#3178: Michael Abshoff: add 64 bit OSX build support to freetype
#3179: Michael Abshoff: more 64 bit OSX libpng fixes
#3181: Michael Abshoff: add 64 bit OSX build support to iml
#3182: Michael Abshoff: improve 64 bit OSX build support for givaro
#3183: Michael Abshoff: add 64 bit OSX build support to linbox
#3186: Michael Abshoff: fix 64 bit OSX build support for numpy
#3187: Michael Abshoff: fix 64 bit OSX build support for matplotlib
#3188: Michael Abshoff: add 64 bit OSX build support to mpfi
#3189: Michael Abshoff: add 64 bit OSX build support to pycrypto
#3190: Michael Abshoff: add 64 bit OSX build support to zodb
#3191: Michael Abshoff: add 64 bit OSX build support to quaddouble
#3192: Michael Abshoff: fix 64 bit OSX build support for python_gnutls
#3197: Michael Abshoff: fix 64 bit OSX build support for m4ri
#3198: Michael Abshoff: fix 64 bit OSX build support for ecm
#3200: Michael Abshoff: fix 64 bit OSX build support for genus2reduction
#3213: Timothy Clemans: notebook -- Account Settings page for changing password and e-mail address
#3220: William Stein: readline -- fix a couple of issues
#3222: William Stein: sqlite -- add cygwin support to sqlite
#3233: William Stein: cygwin -- make linbox work with cygwin
#3224: Michael Abshoff: add 64 bit OSX build support for lcalc
#3225: Michael Abshoff: add 64 bit OSX build support for cddlib
#3226: Michael Abshoff: add 64 bit OSX build support for gfan
#3234: William Stein: cygwin -- make numpy work with cygwin
#3235: William Stein, Michael Abshoff: cygwin -- mpfi; get it to work with Cygwin by fixing configure.ac
#3236: William Stein, Michael Abshoff: cygwin -- get quaddouble to work with cygwin
#3238: William Stein, Michael Abshoff: libfpll spkg -- update to work with cygwin
#3239: William Stein, Michael Abshoff: cygwin polybori -- add Cygwin build support for polybori
#3230: William Stein: cygwin -- new givaro spkg that works around stupidity in cygwin
#3241: William Stein, Michael Abshoff: cygwin -- new rubiks spkg that builds on cygwin
#3243: William Stein: cygwin -- get log2 to work on cygwin
#3246: William Stein: cygwin -- fix broken gsl.ode setup.py code
#3252: Yi Qiang: add kbase functionality to libsingular

Friday, May 16, 2008

First sign of 64 bit OSX 10.5 support in Sage 3.0.x

Hello folks,

64 bit OSX 10.5 support has been promised for a while, but I finally managed a couple hours ago to fix a segfault issue in libSingular that made Sage segfault instantly when any multivariate polynomial ring was created. If you check out the 64 bit OSX port page in the wiki you will see that there are currently 35+ ticket listed which need to be fixed. I have patches for all of them, but the quality is uneven and a couple of those patches need to be redone. So it will take a couple more weeks until all those issues are sorted out and polished patches are merged into the official tree.

I have put together a binary [OSX 10.5 x86-64 only]. It is rather hefty at 0.5GB, but I have left a substantial number of hand build spkgs in the dmg more or less by accident. If you plan to build a 64 bit build of Sage on your OSX machine you should wait a couple more days until 3.0.2 is released which will contain the vast majority of fixes. It is even likely that there will be an official 64 bit OSX 10.5 binary, but currently there are 24 doctest failures. Some are trivial to fix, others are caused by the notebook not working due to _ctypes being broken. There are other small issues with pexpect and a couple additional small issues, but overall we are in pretty good shape. On exit OSX also complains about a couple "Non-aligned pointer being freed", but that seems to be mostly cosmetic and will be fixed soon since I got a good idea what is causing this. We will hopefully be at a point where we can call OSX in 64 bit mode fully supported soon, but there will be some more work involved before we are done.

Cheers,

Michael

Sunday, May 11, 2008

Sage 3.0.2.alpha0 release

Hello folks,

this is Sage 3.0.2.alpha0. What happened? It seems that people were busy and until this morning there wasn't a whole lot to merge. But I had a busy day today and finally these is something to put out. We are still mostly on bug fix only mode, so no big surprises. "sage -sdist" seems to have been broken by David Joyner's #3046, so sage-banner is emtpy [see #3161]. I fixed this in the tarball and it will be fixed in Sage 3.0.2.alpha1.

Binaries and sources in the usual place:

I also created md5sums. There are 70+ tickets with patches in trac, among them quite a number of notebook patches. So it would be very nice if people can do some serious review. Note that the notebook patches have some rather large dependency tree available in the wiki here.

Cheers,

Michael

Merged in alpha0:

#336: William Stein, Timothy Clemans: Create an option to clear all cell output
#905: Burcin Erocal, Michael Abshoff: update ipython to 0.8.2
#1230: William Stein: Quit worksheet behaviour
#1557: William Stein: notebook -- usability improvement after uploading file
#2684: Jason Grout: vertices should not default to red
#2768: Jason Grout: add comparison operators to the fast_float mechanism
#2926: Timothy Clemans: notebook -- Minimalistic change password page for notebook user
#2983: Michael Abshoff: Itanium (RHEL 5) -- singular interface problems in matrix_group.py
#3008: William Stein: first cell in notebook is undeletable
#3020: John Cremona, Martin Albrecht: Speed up Finite Fields of characteristic 2 constructors
#3026: Bjarke Hammersholt Roune: multivariate polynomial rings with no variables do not print properly
#3028: Bjarke Hammersholt Roune: Ideals in multivariate polynomial rings with no variables raise exception on comparison
#3065: Didier Deshommes: empty matrices: frobenius() throws RuntimeError
#3105: Francis Clark: new _latex_ and modified __repr__ for elements of relative number fields
#3109: William Stein: elliptic curves -- implement P.divide(n) for P a point on an elliptic curve and n an integer
#3110: Gary Furnish: fix pbuild dependency bug
#3116: Mike Hansen: 1x1 symbolic matrices don't work right
#3121: Jason Grout, William Stein: @interact grid control
#3125: Robert Miller: chromatic_polynomial incorrectly blocks control-c
#3126: Robert Bradshaw: Cython annotation has unicode errors (e.g. from the notebook)
#3129: Bjarke Hammersholt Roune: The singular interface should not claim to support polynomial rings with no variables
#3136: William Stein: the readme for osx should be changed to delete the line about inotebook()
#3138: Bjarke Roune: Singular multivariate polynomial ring has redundant _repr_ method
#3142: Martin Albrecht: MPolynomialIdeal.homogenize bugfix
#3143: Martin Albrecht: remove references to "/home/was"
#3046: David Joyner: version option returning clone branch name
#3150: Carlo Hamalainen: Memory leak in dancing_links.pyx
#3157: Gary Furnish: Executable target for pbuild
#3158: Michael Abshoff: singular-3-0-4-2-20080405.p1 requires flex
#3159: Tim Abbott, Francois Bissey, Michael Abshoff: Patch adding soname to ntl shared library