Mitesh Patel and I have released 4.5.2.alpha1. Source tarball is at
http://sage.math.washington.edu/home/release/sage-4.5.2.alpha1/sage-4.5.2.alpha1.tar
Upgrade path:
http://sage.math.washington.edu/home/release/sage-4.5.2.alpha1/sage-4.5.2.alpha1/
This release has been tested on sage.math and bsd.math. A build on
t2.math is underway. Please build and test.
This release is now pretty much in feature freeze. We don't plan on
merging any tickets unless they fix blocker issues.
* Known issues
This release includes two tiny unreviewed change: I merged #5396, which
makes lcalc a build prerequisite for the Sage library. Rather than leave
things as they were and have everyone's build fail, I changed
spkg/standard/deps. Someone should test that lcalc really is a build
prereq for the library by removing line 442 of spkg/standard/deps and
trying to build alpha1; it should fail with an error as seen in
http://trac.sagemath.org/sage_trac/ticket/5396#comment:86. Someone
should test that.
The second unreviewed change is the deletion of several extra files in
spkg/standard, as I mentioned in
https://groups.google.com/group/sage-release/t/b6fd67d4d4543129. In the
unlikely case that those files were important or necessary, we can just
copy them from the alpha0 tarball.
Blocker ticket #9582 has not been fixed, so there will be at least one
doctest failure on OS X.
There are a couple Sphinx warnings from the Matlab interface. A ticket
for those will be opened shortly.
Mitesh saw a doctest failure in doc/en/tutorial/latex.rst involving
"\usepackage{tikz}", so please confirm if you see that.
Merged in sagenb:
#8369: Mitesh Patel: Mistake in text on data file upload page [Reviewed by Robert Mařík; merged in sagenb-0.8]
#9554: Leif Leonhardy: Doctest failure in SageNB's sageinspect.py with #8988 [Reviewed by Volker Braun; merged in sagenb-0.8.2]
#9580: Tim Dumol: Check in, ignore, or delete sagenb.po in SageNB package [Reviewed by Mitesh Patel; merged in sagenb-0.8.2]
We merged 83 tickets in sage-4.5.2.alpha1:
#2119: Mike Hansen: matlab matrix conversion issue [Reviewed by Ross Kyprianou]
#3342: Willem Jan Palenstijn: bizarre source code introspection output [Reviewed by John Palmieri]
#5396: Rishikesh, Yann-Laigle Chapuy: Wrapping lcalc library [Reviewed by John Cremona, David Kirkby, Alex Ghitza]
#6186: Craig Citro: two probably-easy-to-fix scope bugs [Reviewed by Carl Witty]
#8156: Paul Zimmermann, Mitesh Patel: new function readdata [Reviewed by Tim Dumol, John H Palmieri, Paul Zimmermann]
#8413: Florent Hivert: Add "Unknown" truth value [Reviewed by Robert Bradshaw]
#8641: Dan Drake, John Palmieri, Mitesh Patel: "sage -t" should exit with nonzero exit code if doctests fail [Reviewed by Willem Jan Palenstijn]
#8976: Leif Leonhardy: squarefree_part() fails on Python types [Reviewed by Robert Miller]
#8993: Simon King: Representation of polynomial quotient rings in Singular [Reviewed by Martin Albrecht]
#9002: Karl-Dieter Crisman: Noise on PPC Mac in parametric_surface.pyx [Reviewed by David Kirkby]
#9027: Rob Beezer: Explain Sage and tex interactions in the tutorial [Reviewed by John Palmieri]
#9066: Karl-Dieter Crisman: Improve documentation in shapes2.py [Reviewed by Minh Van Nguyen]
#9083: Dan Drake, John Palmieri: 'make distclean' fails to remove some files or directories. [Reviewed by Mitesh Patel]
#9222: Alex Ghitza: improve doctest coverage of databases/conway.py [Reviewed by David Loeffler]
#9223: Alex Ghitza, John Cremona: improve doctest coverage of databases/cremona.py [Reviewed by John Cremona, Alex Ghitza]
#9226: David Kirkby: README.txt says " Sage builds with GCC >= 3.x" but it does NOT [Reviewed by Robert Bradshaw]
#9242: David Loeffler: Add docstrings and tests for sage/rings/ideal_monoid.py [Reviewed by Alex Ghitza]
#9243: Dan Drake, Willem Jan Palenstijn: sage-doctest should use powers of 2 for return codes [Reviewed by Willem Jan Palenstijn, Mitesh Patel]
#9251: Florent Hivert: Lazy attribute does not properly handles the doc of Cython methods [Reviewed by Robert Bradshaw]
#9278: Alex Ghitza: remove databases/kohel.py since it is not used and broken [Reviewed by Robert Miller]
#9279: Alex Ghitza: remove databases/tables.py since superseded by newer code [Reviewed by Robert Miller]
#9316: Willem Jan Palenstijn: Spurious (?) "# File not found" error at end of doctests [Reviewed by Mitesh Patel]
#9377: Kwankyu Lee: unable to coerce matrix over finite field into magma [Reviewed by Mariah Lenox]
#9398: Nils Bruin: Sage meddles with soft rlimits [Reviewed by William Stein]
#9456: John Palmieri: zlib should be a prerequisite for libpng [Reviewed by David Kirkby]
#9462: Willem Jan Palenstijn: warning in matrix_modn_dense.pyx [Reviewed by John Palmieri]
#9501: William Stein: Make an @fork decorator [Reviewed by Martin Albrecht]
#9527: William Stein: improve "sage -sh" prompt so it doesn't confuse everybody [Reviewed by Martin Albrecht]
#9561: Karl-Dieter Crisman: Docbuild warnings caused by #9219 [Reviewed by John Palmieri]
#9566: Fredrik Johansson: Allow sage.libs.mpmath.call(..., parent=something) [Reviewed by Harald Schilly]
#9570: Nathann Cohen, Leonardo Sampaio: Wrong LP solver ordering [Reviewed by Nathann Cohen, Leonardo Sampaio]
#9571: Nathann Cohen: Sphinx Warning: Missing title for sage.misc.exceptions after #9249 [Reviewed by John Palmieri]
#9572: Mitesh Patel: SageNB 0.8.2 [Reviewed by Carl Witty]
#9573: Andrey Novoseltsev: Error building the PDF reference manual [Reviewed by John Palmieri]
#9574: Dan Drake: Ignore zope-testrunner in the scripts repository [Reviewed by Mitesh Patel]
#9579: Nathan Cohen: Raise an exception when arguments to add_constraint are not admissible [Reviewed by Harald Schilly]
#9583: Dan Drake: Unhandled SIGSEGV with 4.5.2.alpha0 on t2 [Reviewed by Mitesh Patel]
#9584: Leif Leonhardy: Weird timeouts in doctesting generic_graph with 4.5.2.alpha0 on some systems [Reviewed by John Palmieri]
#9588: Armin Straub: Extend is_prime_power to negative exponents [Reviewed by Carl Witty]
#9589: Alexandre Blondin Masse: Doctest failures in nfactor_enumerable_word.py on 32-bit Linux [Reviewed by Leif Leonhardy]
#9590: Andrey Novoseltsev: Doctest failures in cone.py and toric_lattice_element.pyx on 32-bit Linux [Reviewed by Leif Leonhardy]
#9594: Leif Leonhardy: Spring layout for graphs is currently random across platforms: mark the doctest accordingly [Reviewed by John Palmieri]
#9597: Mitesh Patel: Fix first line of pari-2.3.5.p1's spkg-install [Reviewed by Dan Drake]
--
--- Dan Drake
----- http://mathsci.kaist.ac.kr/~drake
-------
Unless someone objects, this is part of
http://trac.sagemath.org/sage_trac/ticket/5396
> The second unreviewed change is the deletion of several extra files in
> spkg/standard, as I mentioned in
> https://groups.google.com/group/sage-release/t/b6fd67d4d4543129. In the
> unlikely case that those files were important or necessary, we can just
> copy them from the alpha0 tarball.
This is now
http://trac.sagemath.org/sage_trac/ticket/9609
> There are a couple Sphinx warnings from the Matlab interface. A ticket
> for those will be opened shortly.
This is now
http://trac.sagemath.org/sage_trac/ticket/9608
> Mitesh saw a doctest failure in doc/en/tutorial/latex.rst involving
> "\usepackage{tikz}", so please confirm if you see that.
This is now
http://trac.sagemath.org/sage_trac/ticket/9607
> Merged in sagenb:
By the way, we merged a new SageNB package chiefly to fix a doctest
(#9554) and to merge #3342's sage and sagenb repository patches into the
same Sage release.
> This release includes two tiny unreviewed change: I merged #5396, which
> makes lcalc a build prerequisite for the Sage library. Rather than leave
> things as they were and have everyone's build fail, I changed
> spkg/standard/deps. Someone should test that lcalc really is a build
> prereq for the library by removing line 442 of spkg/standard/deps and
> trying to build alpha1; it should fail with an error as seen in
> http://trac.sagemath.org/sage_trac/ticket/5396#comment:86. Someone
> should test that.
Dan,
that's risky. If the builds are performed in parallel, the order of builds would
not be fixed, so what works for 99 people, might fail for the 100th.
The best way to solve it is to find out for sure whether lcalc is a prerequisite
or not by analysis of the source code - not trial and error.
On my Sun Ultra 27 (a relatively fast machine), lcalc takes only 12 seconds to
build, so I doubt making it an unnecessary prerequisite would have a major
impact on the build time of Sage. The Sage library takes about 5 minutes to
build on here, and packages like Singular take over 9 minutes, so I don't think
this would slow down parallel builds significantly.
Dave
That's known: #9607.
> sage -t "devel/sage/sage/libs/lcalc/
> lcalc_Lfunction.pyx"
Haven't seen that before. What platform did you build this on?
Dan
Ubuntu 9.04 x86_64 (Core2, gcc 4.5.0, native code):
make build: OK (parallel build with 12 jobs)
make doc: OK (except the already reported 2 warnings and *)
make testlong: OK (all tests passed with #9607 applied)
Builds/tests on two 32-bit systems in progress...
Btw, thanks for managing this release, well done so far. :-)
-Leif
_____________
* As reported many times before (I haven't opened a ticket for that
though), there's still a conf.py file missing for thematic tutorials.
(This gives a Sphinx ERROR, that Sage simply ignores - I've opened a
ticket for *that* some time ago: #9426, but haven't yet had the time to
fix it.)
Ubuntu 9.04 x86_64 (Core2):
100 runs, all passed, no files left around afterwards.
Perhaps John's ptestlong did leave one [locked?]?
-Leif
I think that's a real bug that pops up when testing on a slow/nfs
filesystem. Martin Albrecht saw it once when refereeing the patch.
The patch should be bounced until I find a way to correctly deal with
this problem.
William
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
I've rerun the test 20 times with a timeout that always "hits"; at least
this doesn't leave any files (nor orphans) around. I'll try later with
DOT_SAGE pointing to some NFS-mounted filesystem...
Btw, we have uses of $DOT_SAGE/tmp, $DOT_SAGE/tmp/tmp and $DOT_SAGE/temp
(perhaps even others, too).
-Leif
Yes. The referee specifically wanted an example of a segfault.
@fork is a cool-arse new decorator we wrote that lets you do stuff
like:
@fork
def crazy_function(...):
crazy stuff that segfaults, leaks memory, etc.
z = crazy_function(...)
and happily go along without any serious trouble.
William
...
>
> Blocker ticket #9582 has not been fixed, so there will be at least one
> doctest failure on OS X.
The following tests failed:
sage -t "devel/sage/doc/en/tutorial/latex.rst"
sage -t "devel/sage/sage/symbolic/random_tests.py"
Total time for all tests: 6966.2 seconds
This is on a power mac running 10.6.4.
>
...
That one is most probably fixed at #9607 (already positive review).
The ticket for those lcalc doctest failures is #9615. It's a blocker for
4.5.2.
Dan
This is now #9616. William, you mentioned that we should just yank the
patch from #9501; is that what you would like for 4.5.2?
As a note to anyone who would like to do that, you can use
hg backout --merge 21d255f024a0
to create a changeset that represents the inverse of the patch at #9501.
Dan
> skynet machine mark (sparc solaris) is still building: it's been more
> than 8 hours now. We'll have a race: see if you can release rc1
> before mark finishes building and doctesting alpha1. On the bright
> side, on this solaris machine with the new gcc compiler built to use
> the sun linker, it's gotten farther in the build than I've ever seen.
That's good news, though I think later you said on a ticket it failed at cvxopt.
It should be possible to build on mark. I've built Sage on a range of SPARCs
* Netra T1
* Blade 1000
* Blade 2000
* T5240
I'm sure we can add Blade 2500 to that list!
> skynet machine fulvia (solaris on x86): the build failed. I think gcc
> is misconfigured.
Was this a 32-bit or 64-bit build?
I've managed to build Sage on OpenSolaris on x86 hardware as a 64-bit binary,
but it crashes at startup. But it needs a few patches - not all of which I have
put on tickets. I hit problems earlier in the build process when I first tried a
32-bit build, and decided it was less hassle and more worthwhile to just
concentrate on a 64-bit build. So I have not tried 32-bit much on x86
> John
>
Dave
This one is fixed at #9607.
http://trac.sagemath.org/sage_trac/ticket/9607
> sage -t -long "devel/sage/sage/interfaces/ecm.py" # Time out
You can increase the timeouts (in seconds) by e.g.
$ export SAGE_TIMEOUT=900 # 15 minutes, for testing without "-long"
$ export SAGE_TIMEOUT_LONG=3600 # 1 hour, for doctesting with "-long"
(The timeouts are wall/real time allowed for each *file* being tested,
not CPU/user time.)
> sage -t -long "devel/sage/sage/symbolic/random_tests.py"
This is most probably #9582, which has not been fixed yet.
http://trac.sagemath.org/sage_trac/ticket/9582
> Total time for all tests: 17431.1 seconds
Wow! ;-)
-Leif