Releasing sage-5.2

[Jeroen Demeyer]

Sage 5.2 is now ready to be released.

Sources:
http://boxen.math.washington.edu/home/release/sage-5.2/

Binaries:
http://boxen.math.washington.edu/home/buildbot/binaries/sage/5.2/

Webmasters: please do the necessary.

Can somebody also bump the Trac milestone 5.2 to 5.3 for open tickets.

[Harald Schilly]

Nice! Sources + 9 new Binaries are on their way to the mirrors in <10
minutes :-)

harald

[Minh Nguyen]

On Mon, Jul 30, 2012 at 6:29 PM, Jeroen Demeyer <jdem...@cage.ugent.be> wrote:
> Webmasters: please do the necessary.

The documentation has been updated.

> Can somebody also bump the Trac milestone 5.2 to 5.3 for open tickets.

Done.

--
Regards,
Minh Van Nguyen
http://bit.ly/mvngu

[leif]

Harald Schilly wrote:
> Nice! Sources + 9 new Binaries are on their way to the mirrors in <10
> minutes :-)

Unfortunately at least all Linux binaries still (or again) have the
broken graphs.db symlink:

/home/sagemath/sage-mirror/linux/32bit/sage-5.2-linux-32bit-ubuntu_12.04_lts-i686-Linux.tar.lzma
/home/sagemath/sage-mirror/linux/32bit/sage-5.0.1-linux-32bit-fedora_release_16_verne_-i686-Linux.tar.lzma
/home/sagemath/sage-mirror/linux/itanium/sage-5.0.1-linux-64bit-red_hat_enterprise_linux_server_release_5.3_tikanga_-ia64-Linux.tar.gz
/home/sagemath/sage-mirror/linux/64bit/sage-5.0.1-linux-64bit-fedora_release_16_verne_-x86_64-Linux.tar.lzma
/home/sagemath/sage-mirror/linux/64bit/sage-5.2-linux-64bit-ubuntu_10.04.3_lts-x86_64-Linux.tar.lzma
/home/sagemath/sage-mirror/linux/64bit/sage-5.2-linux-64bit-red_hat_enterprise_linux_server_release_5.6_tikanga-x86_64-Linux.tar.gz
/home/sagemath/sage-mirror/linux/64bit/sage-5.2-linux-64bit-ubuntu_12.04_lts-x86_64-Linux.tar.lzma
/home/sagemath/sage-mirror/linux/64bit/sage-5.2-linux-64bit-ubuntu_8.04.4_lts-x86_64-Linux.tar.gz


I've made #13292 [1] a blocker now.


-leif


[1] http://trac.sagemath.org/sage_trac/ticket/13292

> On Mon, Jul 30, 2012 at 10:29 AM, Jeroen Demeyer <jdem...@cage.ugent.be> wrote:
>> Sage 5.2 is now ready to be released.
>>
>> Sources:
>> http://boxen.math.washington.edu/home/release/sage-5.2/
>>
>> Binaries:
>> http://boxen.math.washington.edu/home/buildbot/binaries/sage/5.2/
>>
>> Webmasters: please do the necessary.
>>
>> Can somebody also bump the Trac milestone 5.2 to 5.3 for open tickets.
>

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

[Jean-Pierre Flori]

Sage built succesfully, the doc as well (that was horribly long), and passed "make test" on Ubuntu 12.04, 64 bits, using system ATLAS lib.
I tried the new notebook, but when I stopped the server, I got a segfault:
2012-07-30 17:35:42+0200 [-] Server Shut Down.
/media/local/flori/sage/sage-5.2/local/lib/libcsage.so(print_backtrace+0x31)[0x7f91f6f5816a]
/media/local/flori/sage/sage-5.2/local/lib/libcsage.so(sigdie+0x14)[0x7f91f6f5819c]
/media/local/flori/sage/sage-5.2/local/lib/libcsage.so(sage_signal_handler+0x216)[0x7f91f6f57d56]
/lib/x86_64-linux-gnu/libpthread.so.0(+0xfcb0)[0x7f91fd05dcb0]
/lib/x86_64-linux-gnu/libc.so.6(cfree+0x1c)[0x7f91fcd1402c]
/media/local/flori/sage/sage-5.2/local/lib/libcsage.so(+0xa2a9)[0x7f91f6f582a9]
/media/local/flori/sage/sage-5.2/local/lib/libcsage.so(sage_mpir_free+0x1c)[0x7f91f6f58355]
/media/local/flori/sage/sage-5.2/local/lib/libgivaro.so.0(_ZN7IntegerD2Ev+0x18)[0x7f91e4105afc]
/media/local/flori/sage/sage-5.2/local/lib/libgivaro.so.0(_ZN8RationalD2Ev+0x1d)[0x7f91e411009d]
/lib/x86_64-linux-gnu/libc.so.6(+0x3b921)[0x7f91fcccc921]
/lib/x86_64-linux-gnu/libc.so.6(+0x3b9a5)[0x7f91fcccc9a5]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf4)[0x7f91fccb2774]
python[0x400661]

------------------------------------------------------------------------
Unhandled SIGSEGV: A segmentation fault occurred in Sage.
This probably occurred because a *compiled* component of Sage has a bug
in it and is not properly wrapped with sig_on(), sig_off(). You might
want to run Sage under gdb with 'sage -gdb' to debug this.
Sage will now terminate.
------------------------------------------------------------------------
Segmentation fault (core dumped)
True
sage:

I was trying the spkg for FLINT 2.3, so I could have broken sthg, but then I tried to get back a vanilla system by reinstalling the old FLINT spkg, popping the patches related to FLINT 2.3 and rebuilding Sage, but got the same segfault.

[Volker Braun]

Since it seems to be a problem involving libgivaro maybe you can try the new spkg at http://trac.sagemath.org/sage_trac/ticket/9511

[leif]

Just read someone else is having the same issue on Ubuntu 12.04 x86_64
(although using a pre-built binary).

I may be wrong, but to me this smells like another Ubuntu GCC bug (as so
far nobody has reported this error for a different system)...

[Which version do you have, and is it up-to-date?]

Could you try building Sage from scratch with SAGE_INSTALL_GCC=yes (or
some other non-Ubuntu GCC) and see whether the problem remains?


-leif

> I was trying the spkg for FLINT 2.3, so I could have broken sthg, but
> then I tried to get back a vanilla system by reinstalling the old FLINT
> spkg, popping the patches related to FLINT 2.3 and rebuilding Sage, but
> got the same segfault.

[Jeroen Demeyer]

On 2012-08-01 13:32, leif wrote:
> Could you try building Sage from scratch with SAGE_INSTALL_GCC=yes (or
> some other non-Ubuntu GCC) and see whether the problem remains?

That doesn't work as any non-Ubuntu GCC won't build on Ubuntu 12.04.

[leif]

:-)

I did build and use FSF GCC 4.6.3 as well as 4.7.0 on Precise, although
with a beta. Just needed some "manual intervention" (such as creating
symlinks) IIRC; don't recall how much of a headache it was...


-leif

[Jean-Pierre Flori]

I could already try on Debian, but I don't know for sure that Ubuntu adds custom patches on top of that of Debian.

I'll also give headache Sage gcc a try on Precise.

[leif]

Jean-Pierre Flori wrote:
> I could already try on Debian, but I don't know for sure that Ubuntu
> adds custom patches on top of that of Debian.

I doubt Debian uses Ubuntu's GCC (which also contains some upstream FSF
patches from trunk).

> I'll also give headache Sage gcc a try on Precise.

I'd rather try to build it outside of Sage... ;-)

IIRC the biggest headache was GCC's 'make clean' bailing out... :-)

[Keshav Kini]

Daniel Smertnig <daniel....@gmail.com> writes:
> The real question is then of course why the OpenSSL import fails on
> Ubuntu/Debian: The symbol SSLv2_method is undefined in
> libssl.so.1.0.0. As far as I know, this was removed for security
> reasons. Recent versions of PyOpenSSL seem to take this into account
> (by checking OPENSSL_NO_SSL2), so updating PyOpenSSL (in sagenb?)
> should remove the crash and make SSL available on Debian/Ubuntu.

As I recall, we use pyOpenSSL 0.12 (rather than 0.13) intentionally
because 0.13 breaks when the system's OpenSSL is 0.9.x. See
http://trac.sagemath.org/sage_trac/ticket/11080#comment:159 and the
subsequent few comments.

As I commented on #13121, I would certainly like to be able to use the
latest version of pyOpenSSL, though.

-Keshav

----
Join us in #sagemath on irc.freenode.net !

[leif]

Haven't read all of these fairly long threads, but will a later version
of pyOpenSSL again also support older versions of libssl (i.e., 0.9.8+)?


-leif

> Join us in #sagemath on irc.freenode.net !

[Keshav Kini]

leif <not.r...@online.de> writes:
> Keshav Kini wrote:
>> Daniel Smertnig <daniel....@gmail.com> writes:
>>> The real question is then of course why the OpenSSL import fails on
>>> Ubuntu/Debian: The symbol SSLv2_method is undefined in
>>> libssl.so.1.0.0. As far as I know, this was removed for security
>>> reasons. Recent versions of PyOpenSSL seem to take this into account
>>> (by checking OPENSSL_NO_SSL2), so updating PyOpenSSL (in sagenb?)
>>> should remove the crash and make SSL available on Debian/Ubuntu.
>>
>> As I recall, we use pyOpenSSL 0.12 (rather than 0.13) intentionally
>> because 0.13 breaks when the system's OpenSSL is 0.9.x. See
>> http://trac.sagemath.org/sage_trac/ticket/11080#comment:159 and the
>> subsequent few comments.
>>
>> As I commented on #13121, I would certainly like to be able to use the
>> latest version of pyOpenSSL, though.
>
> Haven't read all of these fairly long threads, but will a later
> version of pyOpenSSL again also support older versions of libssl
> (i.e., 0.9.8+)?

Well, that's exactly what I meant by "I would certainly like to be able
to use the latest version of pyOpenSSL". The current release of
pyOpenSSL (0.13) fails to build against OpenSSL 0.9.8e and older. Daniel
found the upstream bug report for this:
https://bugs.launchpad.net/pyopenssl/+bug/845445

I posted a comment confirming that the patch seen on the ticket works on
CentOS 5. I presume it should work on old versions of OS X, the OS in
which *we* first saw this problem, as well. Hopefully the patch is
merged and a new release is made soon. If not, I suppose we could make
an SPKG for pyOpenSSL 0.13 incorporating the patch...

-Keshav

----

[leif]

Ah, thanks.

It wasn't clear to me whether pyOpenSSL (0.13) /intentionally/ broke
support for older libssl versions (e.g. prior to 1.0).


-leif