sage-5.0.beta10 released
Jeroen Demeyer
We're releasing Sage 5.0.beta10.
Source archive:
http://boxen.math.washington.edu/home/release/sage-5.0.beta10/sage-5.0.beta10.tar
Upgrade path:
http://boxen.math.washington.edu/home/release/sage-5.0.beta10/sage-5.0.beta10/
The source and upgrade path can also be found on the mirror network
(you might need to wait a while before the mirrors are synchronized):
http://www.sagemath.org/download-latest.html
Please build, test, and report! We'd love to hear about your
experiences with this release.
== Tickets ==
* We closed 359 tickets in this release. For details, see
http://boxen.math.washington.edu/home/release/sage-5.0.beta10/tickets.html
Closed tickets:
#1386: Implement splitting fields [Reviewed by Jeroen Demeyer]
#3517: "sage -upgrade" does not upgrade $SAGE_ROOT/sage [Reviewed by
Jeroen Demeyer]
#3898: [with spkg, needs work] Make an optional, self contained gcc
4.3.4 spkg [Reviewed by Jonathan Bober, David Roe]
#6365: bug in constructing extensions of finite fields [Reviewed by Jen
Balakrishnan]
#7066: sympow ignores CC and uses gcc even when CC is set to Sun's
compiler [Reviewed by R. Andrew Ohana]
#12184: Change example in doc/bordeaux/elliptic_curves.rst [Reviewed by
R. Andrew Ohana]
#12712: John Palmieri: update optional spkg 'database_gap' [Reviewed by
Jeroen Demeyer, Dmitrii Pasechnik]
Merged in sage-5.0.beta10:
#6588: Nicolas M. Thiéry: Categories for root systems and many misc
improvements [Reviewed by Anne Schilling, Mark Shimozono, Florent Hivert]
#10492: Jeroen Demeyer: Rework spkg/install, spkg/standard/deps,
spkg/standard/newest_version [Reviewed by John Palmieri]
#11585: Luca De Feo, Florent Hivert: Make deprecated_function_alias
print the whole module path when it differs from the original [Reviewed
by Rob Beezer]
#11586: David Roe: bug in p-adic extension norm method [Reviewed by
Jennifer Balakrishnan]
#11615: John Palmieri: optional doctests: clean up and document "known
bug", "optional: requires PKG" [Reviewed by Karl-Dieter Crisman]
#11634: Volker Braun: Base Polyhedron on PPL (by default) [Reviewed by
Marshall Hampton, David Loeffler, Andrey Novoseltsev]
#11921: Peter Story, Burcin Erocal: Allow Sympy conversion sign/sgn
[Reviewed by Karl-Dieter Crisman]
#12018: R. Andrew Ohana: sage-list-packages will fail if user can't
write to SAGE_ROOT [Reviewed by John Palmieri]
#12062: David Roe: FiniteField_ntl_gf2e to Python [Reviewed by David
Loeffler]
#12355: Nathann Cohen: Bug in Graph.girth [Reviewed by David Coudert,
Jeroen Demeyer]
#12499: Punarbasu Purkayastha: The density() function of a sparse matrix
looks at every matrix entry [Reviewed by Rob Beezer]
#12512: Vincent Delecroix: fixed point and periodic point of word
morphism [Reviewed by Timo Jolivet]
#12541: Andrey Novoseltsev: Remove Sequence test in span [Reviewed by
Rob Beezer]
#12602: Jeroen Demeyer: Rework download/extract code in sage-spkg
[Reviewed by John Palmieri]
#12609: John Palmieri, Jeroen Demeyer: Update developer guide for SPKGs
[Reviewed by Jeroen Demeyer, John Palmieri]
#12617: R. Andrew Ohana: Update Cremona's table of elliptic curves
[Reviewed by John Cremona, Jennifer Balakrishnan]
#12640: William Stein: some special cases of dimension_new_cusp_forms
are broken [Reviewed by Jen Balakrishnan]
#12644: R. Andrew Ohana: redundant source in elliptic_curves spkg
[Reviewed by Jennifer Balakrishnan]
#12670: John Palmieri: Store lazy import caches in .sage/cache/ rather
than in .sage [Reviewed by Florent Hivert]
#12678: Nicolas M. Thiéry: Extract a method to get the name of the .rst
file corresponding to a python module, and doctest it. [Reviewed by Mike
Hansen]
#12689: William Stein: Segfault in solve_left for sparse matrices over
ZZ [Reviewed by Douglas McNeil]
#12691: David Roe, William Stein: Create a stopgap warning [Reviewed by
R. Andrew Ohana]
#12694: Keshav Kini: Fix four SPKGs with irregular directory structures
[Reviewed by R. Andrew Ohana]
#12698: Jeroen Demeyer: Add a warning not to install packages using the
Sage shell [Reviewed by John Palmieri]
#12701: Jen Balakrishnan: Stopgap for 6667 [Reviewed by R. Andrew Ohana]
#12702: John Palmieri: document "stopgaps" [Reviewed by R. Andrew Ohana]
William Stein
On Saturday, March 24, 2012, Jeroen Demeyer <jdem...@cage.ugent.be> wrote:
> Dear Sage lovers,
>
> We're releasing Sage 5.0.beta10.
>
> You received this message because you are subscribed to the Google Groups "sage-release" group.
> To post to this group, send email to sage-r...@googlegroups.com.
> To unsubscribe from this group, send email to sage-release...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/sage-release?hl=en.
>
>
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
Anthony David
Dima Pasechnik
On Saturday, 24 March 2012 20:24:07 UTC+8, adavid wrote:
As per beta9:$ lsb_release -d -cDescription: Ubuntu 10.10Codename: maverickexport SAGE_ATLAS_LIB=/usr/libexport SAGE_CHECK="yes"cvxopt test failed:Testing portfolio.py ...Traceback (most recent call last):File "portfolio.py", line 7, in <module>import pylabImportError: No module named pylabError: test /home/adavid/build/sage-5.0.beta10/spkg/build/cvxopt-1.1.3.p1/src/examples/doc/chap8/portfolio.py failedWorkaround:$ ./sage -i matplotlib$ make...Testing portfolio.py ...Testing qcl1.py ...Subsequent building/testing continues
Anthony
To unsubscribe from this group, send email to sage-release+unsubscribe@googlegroups.com.
John H Palmieri
On Saturday, March 24, 2012 4:16:09 AM UTC-7, William Stein wrote:
On Saturday, March 24, 2012, Jeroen Demeyer wrote:
> Dear Sage lovers,
>
> We're releasing Sage 5.0.beta10.
>
I use os x 10.7 currently with xcode 4.2. How can I build this sage version so i can sctually use it? I tried 5.0.beta9, but it crashes on a lot of doctests (though it builds).
Use the tar ball posted at #12369.
On OS X 10.7, I get one doctest failure:
sage -t --long -force_lib devel/sage/sage/matrix/matrix_double_dense.pyx
**********************************************************************
File "/Users/palmieri/Desktop/Sage_stuff/sage_builds/GCC-no-check/sage-5.0.beta10-gcc/devel/sage-
main/sage/matrix/matrix_double_dense.pyx", line 2843:
sage: B.is_normal(algorithm='naive', tol=1.0e-34)
Expected:
False
Got:
True
**********************************************************************
I think this is from #11104.
--
John
Justin C. Walker
On Mar 24, 2012, at 02:20 , Jeroen Demeyer wrote:
> Dear Sage lovers,
>
> We're releasing Sage 5.0.beta10.
>
> Source archive:
>
> http://boxen.math.washington.edu/home/release/sage-5.0.beta10/sage-5.0.beta10.tar
Built from scratch on Mac OS X, 10.6.8 (Dual 6-core Xeons) w/o problems.
All tests ('ptestong') passed!
Justin
--
Justin C. Walker
Curmudgeon at Large
Director
Institute for the Enhancement of the Director's Income
--
Build a man a fire and he'll be warm
for a night.
Set a man on fire and he'll be warm
for the rest of his life.
Justin C. Walker
On Mar 24, 2012, at 07:43 , John H Palmieri wrote:
>
>
> On Saturday, March 24, 2012 4:16:09 AM UTC-7, William Stein wrote:
>>
>>
>>
>> On Saturday, March 24, 2012, Jeroen Demeyer wrote:
>>> Dear Sage lovers,
>>>
>>> We're releasing Sage 5.0.beta10.
>>>
>>
>> I use os x 10.7 currently with xcode 4.2. How can I build this sage
>> version so i can sctually use it? I tried 5.0.beta9, but it crashes on a
>> lot of doctests (though it builds).
>>
>
> Use the tar ball posted at #12369.
Tried this on Mac OS X, 10.7.1 (Core i7). The build (-j7) failed on the MPIR build. The immediate cause is
checking compiler gcc -march=native -g -O3 ... no, double -> ulong conversion
configure: error: could not find a working compiler, see config.log for details
Error configuring MPIR. (See above for the options passed to it.)
FWIW, 23 other spkg's built; this one went boom and the whole build collapsed.
The full log for MPIR is at
sage.math.washington.edu:~justin/logs/mpir-2.1.3.p9.log
Justin
--
Justin C. Walker
Curmudgeon-at-large
Director
Institute for the Absorption of Federal Funds
----
186,000 Miles per Second
Not just a good idea:
it's the law!
----
Justin C. Walker
On Mar 24, 2012, at 18:38 , Justin C. Walker wrote:
>
> On Mar 24, 2012, at 07:43 , John H Palmieri wrote:
>
>>
>>
>> On Saturday, March 24, 2012 4:16:09 AM UTC-7, William Stein wrote:
>>>
>>>
>>>
>>> On Saturday, March 24, 2012, Jeroen Demeyer wrote:
>>>> Dear Sage lovers,
>>>>
>>>> We're releasing Sage 5.0.beta10.
>>>>
>>>
>>> I use os x 10.7 currently with xcode 4.2. How can I build this sage
>>> version so i can sctually use it? I tried 5.0.beta9, but it crashes on a
>>> lot of doctests (though it builds).
>>>
>>
>> Use the tar ball posted at #12369.
>
> Tried this on Mac OS X, 10.7.1 (Core i7). The build (-j7) failed on the MPIR build.
I should mention that I used the full b10-gcc tarball attached to #12369; not the gcc .spkg. I'm trying "-j1" now.
Justin
--
Justin C. Walker, Curmudgeon at Large
Director
Institute for the Enhancement of the Director's Income
-----------
Nobody knows the trouble I've been
-----------
leif
> On Mar 24, 2012, at 07:43 , John H Palmieri wrote:
>> On Saturday, March 24, 2012 4:16:09 AM UTC-7, William Stein wrote:
>>> On Saturday, March 24, 2012, Jeroen Demeyer wrote:
>>>> Dear Sage lovers,
>>>>
>>>> We're releasing Sage 5.0.beta10.
>>>>
>>>
>>> I use os x 10.7 currently with xcode 4.2. How can I build this sage
>>> version so i can sctually use it? I tried 5.0.beta9, but it crashes on a
>>> lot of doctests (though it builds).
>>>
>>
>> Use the tar ball posted at #12369.
>
> Tried this on Mac OS X, 10.7.1 (Core i7). The build (-j7) failed on the MPIR build. The immediate cause is
>
> checking compiler gcc -march=native -g -O3 ... no, double -> ulong conversion
> configure: error: could not find a working compiler, see config.log for details
> Error configuring MPIR. (See above for the options passed to it.)
>
> FWIW, 23 other spkg's built; this one went boom and the whole build collapsed.
>
> The full log for MPIR is at
> sage.math.washington.edu:~justin/logs/mpir-2.1.3.p9.log
Presumably more, or at least also interesting would be the config.log
from the (temporary) build dir --
$SAGE_ROOT/spkg/build/mpir-2.1.3.p9/src/config.log -- since 'configure'
failed.
-leif
--
() The ASCII Ribbon Campaign
/\ Help Cure HTML E-Mail
Justin C. Walker
On Mar 24, 2012, at 18:50 , leif wrote:
> Justin C. Walker wrote:
>> On Mar 24, 2012, at 07:43 , John H Palmieri wrote:
>>> On Saturday, March 24, 2012 4:16:09 AM UTC-7, William Stein wrote:
>>>> On Saturday, March 24, 2012, Jeroen Demeyer wrote:
[snip]
>
> Presumably more, or at least also interesting would be the config.log from the (temporary) build dir -- $SAGE_ROOT/spkg/build/mpir-2.1.3.p9/src/config.log -- since 'configure' failed.
Ask, and it shall be answered. Let me know if you need more details.
leif
:-)
So your assembler doesn't understand the instructions GCC (4.6.3)
generates (with '-march=native'). You can try specifying some other CPU
type (with '-march=...', e.g. 'core2') in CFLAGS, or try using
'-mno-avx', both a bit odd.
Unfortunately there's currently no other way to disable '-march=native'
if your /compiler/ supports it, other than specifying another CPU for
tuning and/or code generation, or by setting SAGE_FAT_BINARY=yes.
Note that this is not an upstream problem, but rather related to how we
configure MPIR in the spkg. (I'm working on newer ones, so can probably
add a better solution to that.)
(MPIR uses a different assembler for its assembly files, so there
shouldn't be a problem when MPIR chooses code optimized for your
specific CPU your native assembler doesn't understand. Only GCC doesn't
use that assembler, so must not generate instructions the latter doesn't
support.)
FWIW, Simon King has just created a binutils-2.22 spkg btw. (which
includes the GNU assembler); no idea whether that would work on MacOS X...
leif
> Justin C. Walker wrote:
>> On Mar 24, 2012, at 18:50 , leif wrote:
>>> Presumably more, or at least also interesting would be the config.log
>>> from the (temporary) build dir --
>>> $SAGE_ROOT/spkg/build/mpir-2.1.3.p9/src/config.log -- since
>>> 'configure' failed.
>>
>> Ask, and it shall be answered. Let me know if you need more details.
>
> :-)
>
> So your assembler doesn't understand the instructions GCC (4.6.3)
> generates (with '-march=native'). You can try specifying some other CPU
> type (with '-march=...', e.g. 'core2') in CFLAGS, or try using
> '-mno-avx', both a bit odd.
>
> Unfortunately there's currently no other way to disable '-march=native'
> if your /compiler/ supports it, other than specifying another CPU for
> tuning and/or code generation, or by setting SAGE_FAT_BINARY=yes.
>
> Note that this is not an upstream problem, but rather related to how we
> configure MPIR in the spkg. (I'm working on newer ones, so can probably
> add a better solution to that.)
P.S.:
Justin case you want to hack the spkg, the "offending" part of MPIR
2.1.3.p9's spkg-install is:
native_build_cflags_if_supported() # Give option for a native build if
supported.
{
# Some GCC 4.0.x versions don't support '-march=native', and it's
currently
# not supported on all platforms supported by Sage.
# (Compilers other than 'gcc' might support it as well, in which
case the
# following also works. Other flavours could be added, too.)
if touch foo.c && $CC -march=native -c foo.c &>/dev/null; then
echo "-march=native"
fi
rm -f foo.* >/dev/null
}
You could replace this by / temporarily change it to e.g.
native_build_cflags_if_supported()
{
echo "" # not really necessary, but easier to understand
return # so you don't have to delete code below... :-)
leif
> file corresponding to a python module, and doctest it. [Reviewed by Mike
> Hansen]
> #12689: William Stein: Segfault in solve_left for sparse matrices over
> ZZ [Reviewed by Douglas McNeil]
> #12691: David Roe, William Stein: Create a stopgap warning [Reviewed by
> R. Andrew Ohana]
> #12694: Keshav Kini: Fix four SPKGs with irregular directory structures
> [Reviewed by R. Andrew Ohana]
> #12698: Jeroen Demeyer: Add a warning not to install packages using the
> Sage shell [Reviewed by John Palmieri]
> #12701: Jen Balakrishnan: Stopgap for 6667 [Reviewed by R. Andrew Ohana]
> #12702: John Palmieri: document "stopgaps" [Reviewed by R. Andrew Ohana]
>
Ubuntu 10.04.4 LTS x86_64 (with a 2.6.38 kernel; AMD Fusion E-450):
I've replaced the following packages due to (partially known) issues
with them, i.e., not respecting CC or CXX (and I use e.g. CC=gcc-4.6.3),
being prepared with too old autotools to support LTO (GLPK only), or
causing trouble with multiple words in LDFLAGS (Maxima):
* GAP (see #7041 [1])
* GLPK (see #12703 [3])
* Maxima (haven't yet created a ticket for that)
* zn_poly (see #12433 [2])
GCC 4.6.3 with LTO (and Graphite loop optimization) enabled*:
* Singular fails to build. This is a GCC (lto1) bug which pops up
during linking (when link-time optimization actually happens).
See #12738 [4]; Simon King already reported this for a modified
version of the new GCC (4.6.3) spkg along with some additional
prerequisite spkgs he created.
* After working around that issue (by disabling actual LTO), ptestlong
passes all test.
GCC 4.7.0 with LTO (and Graphite loop optimization) enabled*:
* ECL fails to build; assembler error during actual LTO of 'ecl_min'.
(Likely to be a GCC bug; haven't yet opened a ticket for that, nor
examined that further.)
* After "successfully" building ECL without actual LTO (by removing
'-flto' from LDFLAGS), the Maxima build (i.e., 'ecl') hangs.
Using my Maxima spkg that (conditionally) unsets LDFLAGS (and uses
'ecl -norc ...') the Maxima build completes without errors.
Don't know yet whether the former (or the latter ;-) ) is reproducible.
* Singular build has passed the point where it failed with GCC 4.6.3
(and LTO taking place).
* Build hasn't yet finished (but so far 58 spkgs have been built
without errors); more to come...
(Both builds performed with binutils 2.22; GCCs built with GMP 5.0.4,
MPFR 3.1.0-p6, MPC 0.9, PPL 0.12, CLooG-PPL 0.15.11, libelf 0.8.13; FWIW.)
-leif
__________
* CFLAGS et al.: -march=native -flto -floop-interchange
-floop-strip-mine -floop-block -fgraphite-identity -O3 -fno-strict-aliasing
LDFLAGS: -flto -fuse-linker-plugin (the latter only necessary if GCC
had been built in presence of a linker that doesn't support plugins,
like GNU ld versions < 2.21)
See also #12703 for some (off-ticket-topic) comments on GCC, LTO etc.
[1] http://trac.sagemath.org/sage_trac/ticket/7041
[2] http://trac.sagemath.org/sage_trac/ticket/12433
[3] http://trac.sagemath.org/sage_trac/ticket/12703
[4] http://trac.sagemath.org/sage_trac/ticket/12738
Keshav Kini
-Keshav
----
Join us in #sagemath on irc.freenode.net !
Jeroen Demeyer
> This release has not propagated to the mirrors, it seems. See
> http://www.sagemath.org/download-latest.html .
My mistake. Shoule be fixed now (well, once the mirrors are synchronized).
Alexander Dreyer
> We're releasing Sage 5.0.beta10.
Regards,
Alexander
--
Dr. rer. nat. Dipl.-Math. Alexander Dreyer
Abteilung "Systemanalyse, Prognose und Regelung"
Fraunhofer Institut f�r Techno- und Wirtschaftsmathematik (ITWM)
Fraunhofer-Platz 1
67663 Kaiserslautern
Telefon +49 (0) 631-31600-4318
Fax +49 (0) 631-31600-1099
E-Mail alexande...@itwm.fraunhofer.de
Internet http://www.itwm.fraunhofer.de/sys/dreyer.html
leif
The Singular build completed successfully.
> * Build hasn't yet finished (but so far 58 spkgs have been built without
> errors); more to come...
Now at 95 packages... ;-)
* gfan fails to build due to a trivial C++ programming error.
I've made a gfan-0.4plus.p2 package fixing this and also cleaning
up 'spkg-install', but haven't yet opened a ticket for that. [The spkg
should presumably get updated anyway, but someone else should do that on
another ticket.]
* PolyBoRi fails to build due to stricter (C++11) scoping / name
look-up rules.
Unfortunately PolyBoRi doesn't honor CXXFLAGS, so I had to use
CXX="g++-4.7.0 -fpermissive" to make the spkg build.
* LinBox fails to build for the same reason, but does respect
CXXFLAGS, so builds with '-fpermissive' in them.
* The Sage library fails to build for the same reason, but this is
really a mess, as it not only ignores CXXFLAGS, but also uses CC (i.e.,
'gcc-4.7.0') to compile some C++ sources (as well as C sources of
course), and '-fpermissive' is not a valid command-line option to 'gcc'.
Furthermore, with './sage -b', after the build of one module has
failed, the script exits and leaves a couple of (busy) orphans, still
compiling some modules, to no avail btw., as the number of modules to
build doesn't decrease further.
When using 'make', the build hangs, with a Python process waiting
for its children, who in turn wait for something forever.
Obviously, I couldn't run Sage's test suite yet.
-leif
Alexander Dreyer
> * PolyBoRi fails to build due to stricter (C++11) scoping / name
> look-up rules.
https://bitbucket.org/brickenstein/polybori/changeset/43b3931ceed7/raw/
> Unfortunately PolyBoRi doesn't honor CXXFLAGS, so I had to use
> CXX="g++-4.7.0 -fpermissive" to make the spkg build.
I could provide a patch for the PolyBoRi 0.8.x packages, that the
configuration reads the flagsfrom the environment.
leif
> Hi Leif,
>> * PolyBoRi fails to build due to stricter (C++11) scoping / name
>> look-up rules.
> There's an upstream patch for this issue:
> https://bitbucket.org/brickenstein/polybori/changeset/43b3931ceed7/raw/
>
>> Unfortunately PolyBoRi doesn't honor CXXFLAGS, so I had to use
>> CXX="g++-4.7.0 -fpermissive" to make the spkg build.
> I could provide a patch for the PolyBoRi 0.8.x packages, that the
> configuration reads the flags from the environment.
Nice, thanks.
Could you provide a PolyBoRi 0.8.0.p2 spkg including both? :P
(Don't know whether there are already tickets upgrading to some later
0.8.x version.)
Cheers,
-leif
Alexander Dreyer
>
> Could you provide a PolyBoRi 0.8.0.p2 spkg including both? :P
>
> (Don't know whether there are already tickets upgrading to some later
> 0.8.x version.)
See also: http://trac.sagemath.org/sage_trac/ticket/12750
There an spkg for PolyBoRi 0.8.1. I'll merge that later, if
polybori-0.8.0.p2.spkg is fine.
Regards,
Alexander
leif
See #12750 [5]. Alexander has already made a PolyBoRi 0.8.0.p2 spkg
fixing this.
> * LinBox fails to build for the same reason, but does respect CXXFLAGS,
> so builds with '-fpermissive' in them.
Anyway, /someone/(tm) should fix the spkg.
> * The Sage library fails to build for the same reason, but this is
> really a mess, as it not only ignores CXXFLAGS, but also uses CC (i.e.,
> 'gcc-4.7.0') to compile some C++ sources (as well as C sources of
> course), and '-fpermissive' is not a valid command-line option to 'gcc'.
Fixed. Turned out that all non-conforming code originated from C++
headers installed by spkgs, namely Givaro and PolyBoRi.
Since Alexander has already fixed PolyBoRi's (see above), I'll then just
make a new Givaro 3.2.13.rc1.p4 spkg with my fixed headers. [Btw.,
despite that the Givaro spkg has been touched a couple of times, it has
last been upgraded in 2008 -- to a release candidate. But I'm not going
to upgrade it either, at least not at the moment.]
> Furthermore, with './sage -b', after the build of one module has failed,
> the script exits and leaves a couple of (busy) orphans, still compiling
> some modules, to no avail btw., as the number of modules to build
> doesn't decrease further.
> When using 'make', the build hangs, with a Python process waiting for
> its children, who in turn wait for something forever.
>
>
> Obviously, I couldn't run Sage's test suite yet.
Now in progress...
-leif
>> (Both builds performed with binutils 2.22; GCCs built with GMP 5.0.4,
>> MPFR 3.1.0-p6, MPC 0.9, PPL 0.12, CLooG-PPL 0.15.11, libelf 0.8.13;
>> FWIW.)
>>
>>
>> -leif
>>
>> __________
>> * CFLAGS et al.: -march=native -flto -floop-interchange
>> -floop-strip-mine -floop-block -fgraphite-identity -O3
>> -fno-strict-aliasing
>> LDFLAGS: -flto -fuse-linker-plugin (the latter only necessary if GCC had
>> been built in presence of a linker that doesn't support plugins, like
>> GNU ld versions < 2.21)
>>
>> See also #12703 for some (off-ticket-topic) comments on GCC, LTO etc.
>>
>>
>> [1] http://trac.sagemath.org/sage_trac/ticket/7041
>>
>> [2] http://trac.sagemath.org/sage_trac/ticket/12433
>>
>> [3] http://trac.sagemath.org/sage_trac/ticket/12703
>>
>> [4] http://trac.sagemath.org/sage_trac/ticket/12738
[5] http://trac.sagemath.org/sage_trac/ticket/12750
leif
"All tests passed!" ('make ptestlong')
This time I agree with the exclamation mark.
tl;dr: GCC 4.7.0 (with LTO and Graphite loop optimization enabled),
Ubuntu 10.04.4 LTS x86_64
Jonathan Bober
on beta 10, on boxen, and it failed. The log is at
This works on beta8.
(Maybe I should send this to sage-devel instead of sage-release to get
someone's attention?)
> --
> You received this message because you are subscribed to the Google Groups "sage-release" group.
> To post to this group, send email to sage-r...@googlegroups.com.
> To unsubscribe from this group, send email to sage-release...@googlegroups.com.
Jeroen Demeyer
> I just tried installing the gap_packages spkg (sage -i gap_packages)
> on beta 10, on boxen, and it failed. The log is at
>
> http://sage.math.washington.edu/home/lmfdbweb/sage-5.0.beta10-boxen-x86_64-Linux/spkg/logs/gap_packages.log
Confirmed.
leif
> We're releasing Sage 5.0.beta10.
Ubuntu 12.04 beta x86_64 (with latest updates; 2nd gen. Core i5):
Ubuntu's GCC (4.6.3-1ubuntu3) crashes on MPFR (cc1) and Singular
(cc1plus) when building with "-O3". "-O2" in contrast works; also MPFR
passes its test suite with that. (Further flags: "-march=native -g").
Already reported upstream (Ubuntu Launchpad); Dan Drake couldn't
reproduce these errors, although in a /32-bit/ VM I think.
ptestlong passed without errors.
-leif
Dan Drake
> Jeroen Demeyer wrote:
> >We're releasing Sage 5.0.beta10.
>
> Ubuntu 12.04 beta x86_64 (with latest updates; 2nd gen. Core i5):
>
> Ubuntu's GCC (4.6.3-1ubuntu3) crashes on MPFR (cc1) and Singular
> (cc1plus) when building with "-O3". "-O2" in contrast works; also
> MPFR passes its test suite with that. (Further flags:
> "-march=native -g").
>
> Already reported upstream (Ubuntu Launchpad); Dan Drake couldn't
> reproduce these errors, although in a /32-bit/ VM I think.
Yes, I have a 32-bit VM of Ubuntu 12.04 beta. I built beta11 with
env CFLAGS="-O3" make
and mpfr built fine. So did Singular (which Leif said might have
problems), although looking in the logs, I saw things like "gcc -O3 -O2
...." so perhaps it wasn't really using O3.
I don't know if the build finished properly (it's on my work computer,
and I'm at home). I'll test it as soon as I can.
Dan
--
--- Dan Drake
----- http://mathsci.kaist.ac.kr/~drake
-------
leif
> On Fri, 30 Mar 2012 at 04:24PM +0200, leif wrote:
>> Jeroen Demeyer wrote:
>>> We're releasing Sage 5.0.beta10.
>>
>> Ubuntu 12.04 beta x86_64 (with latest updates; 2nd gen. Core i5):
>>
>> Ubuntu's GCC (4.6.3-1ubuntu3) crashes on MPFR (cc1) and Singular
>> (cc1plus) when building with "-O3". "-O2" in contrast works; also
>> MPFR passes its test suite with that. (Further flags:
>> "-march=native -g").
>>
>> Already reported upstream (Ubuntu Launchpad); Dan Drake couldn't
>> reproduce these errors, although in a /32-bit/ VM I think.
>
> Yes, I have a 32-bit VM of Ubuntu 12.04 beta. I built beta11 with
>
> env CFLAGS="-O3" make
>
> and mpfr built fine. So did Singular (which Leif said might have
> problems), although looking in the logs, I saw things like "gcc -O3 -O2
> ...." so perhaps it wasn't really using O3.
I think some parts of Singular are forced to be built with "-O2".
But since g++ / cc1plus crashed for me (on Singular), you'd (also) have to
export CXXFLAGS="-O3"
and probably also
export CPPFLAGS="-O3"
to eventually reproduce that.
(The latter are flags for the C preprocessor and should get used as
well; sometimes they're passed /after/ $CFLAGS/$CXXFLAGS and some
developer-chosen ones, such that one can override the latter by putting
e.g. "-O3" into these.)
> I don't know if the build finished properly (it's on my work computer,
> and I'm at home). I'll test it as soon as I can.
You could retry building Singular with
$ env CXXFLAGS="-O3" CPPFLAGS="-O3" ./sage -f spkg/standard/singular-*.spkg
Thanks,
leif
Exactly that is the case for Singular; '-O3' in CXXFLAGS gets
overridden, while '-O3' in CPPFLAGS doesn't.
>> I don't know if the build finished properly (it's on my work computer,
>> and I'm at home). I'll test it as soon as I can.
>
> You could retry building Singular with
>
> $ env CXXFLAGS="-O3" CPPFLAGS="-O3" ./sage -f spkg/standard/singular-*.spkg
So it's (necessary and) sufficient to try
$ env CPPFLAGS="-O3" ./sage -f spkg/standard/singular-3-1-3-3.p6.spkg
(perhaps with in addition "-march=native", but I don't think that makes
any difference w.r.t. the crash here).
Dan Drake
> I think some parts of Singular are forced to be built with "-O2".
>
> But since g++ / cc1plus crashed for me (on Singular), you'd (also) have to
>
> export CXXFLAGS="-O3"
>
> and probably also
>
> export CPPFLAGS="-O3"
>
> to eventually reproduce that.
I used both of those variables (even though you figured out that only
one is necessary) and CFLAGS=-O3, and sure enough, I get an internal
compiler error:
g++ -O3 -O2 -g -fPIC -pipe -fno-implicit-templates -I. -I.. -I/home/foo/sage-5.0.beta11/local -I/home/foo/sage-5.0.beta11/local/include -I/home/foo/sage-5.0.beta11/local/include -I/home/foo/sage-5.0.beta11/local/include -O3 -DNDEBUG -DOM_NDEBUG -Dix86_Linux -DHAVE_CONFIG_H -c sparsmat.cc
sparsmat.cc: In function ‘BOOLEAN _ZL11smIsNegQuotP8spolyrecS0_S0_.part.12(poly, poly, poly)’:
sparsmat.cc:1948:16: internal compiler error: Segmentation fault
This on a 32-bit Ubuntu 12.04 VM with all updates. gcc is
gcc version 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu3).
I'll try mpfr and see what happens there. With only CFLAGS set to O3,
everything built and tests passed.