Re: [eigen] 3.2.10 released!

On Tue, Oct 4, 2016 at 10:59 PM, Benoit Steiner <benoit.steiner.goog@xxxxxxxxx> wrote:

I setup nightly builds for 64 bit ARM some time ago. The results are available on this test dashboard.

Except some failures in unsupported modules everything looks OK for that one.

I also remember testing ppc64 and ppc64le systems very recently with success, I'll rerun them tomorrow.

gael

On Tue, Oct 4, 2016 at 12:53 PM, Thomas Klimpel <jacques.gentzen@xxxxxxxxx> wrote:
The sad truth is that I'm not really helping to test Eigen releases on various compilers, but that those are the compilers I and my colleagues have to use. (Maybe we could avoid it, and we certainly can just continue to use Eigen 3.2.x.)

I think I understand now how booost.mpl/boost.type_traits writes those has_binary_operator, has_unary_operator, and has_nullary_operator traits. (The _expression_ in sizeof calls those operators, and some tricks ensure that an operator operator with "void" return type is found if the correct operator doesn't exist). But I won't find time to (try to) write them before tomorrow afternoon.

On Tue, Oct 4, 2016 at 7:09 PM, Benoit Jacob <jacob.benoit.1@xxxxxxxxxx> wrote:
And thanks to Thomas and everyone who's helping to test Eigen releases on various compilers!

2016-10-04 14:07 GMT-04:00 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
Just a quick bit cheerleading, especially directed to Gael and Christoph: the quality of support that you're offering here is amazing, doing a Eigen 3.2.10 now is a great way to support users ahead of the upcoming 3.3 release, and supporting gcc 4.4 is already very generous already given it's 7.5 years old.

Thanks!!

2016-10-04 13:35 GMT-04:00 Christoph Hertzberg <chtz@xxxxxxxxxxxxxxxxxxxxxxxx>:
Regarding GCC we do only claim to support version 4.4 and newer, so I agree that we should not waste time on that.

Christoph

On 2016-10-04 13:43, Gael Guennebaud wrote:

Thank you for the feedback, it's reassuring that it still works with
MSVC2010 (thanks to your efforts). Regarding gcc 4.1, I would not waste
time on it. It's really too old, and it always gave poor performance with
Eigen.

gael

On Tue, Oct 4, 2016 at 1:30 PM, Thomas Klimpel <jacques.gentzen@xxxxxxxxx>
wrote:

Eigen 3.3 will be released very soon, and we encourage everybody to

already move to the 3.3-rc1 version.

Thanks for the great work. I just tested the default branch with MSVC2010,
gcc 4.8.4, gcc 4.4.6 and gcc 4.1.2. Only gcc 4.1.2 fails:

Eigen/src/Core/util/Meta.h:410: error: no matching function for call to
Eigen::internal::scalar_constant_op<double>::operator()(long int, long
int) const
Eigen/src/Core/util/Meta.h:401: error: no matching function for call to
Eigen::internal::scalar_constant_op<double>::operator()(long int) const
Eigen/src/Core/util/Meta.h:392: error: no matching function for call to
Eigen::internal::scalar_identity_op<double>::operator()() const

I tried to fix it, but the obvious trivial fixes didn't work. I will have
to lookup how boost.mpl (or rather boost/type_traits/details/has:binary_operator.hpp)
does it. At least that is my current plan. I will try to post a patch with
a fix today.

--
Dipl. Inf., Dipl. Math. Christoph Hertzberg

Universität Bremen
FB 3 - Mathematik und Informatik
AG Robotik
Robert-Hooke-Straße 1
28359 Bremen, Germany

Zentrale: +49 421 178 45-6611

Besuchsadresse der Nebengeschäftsstelle:
Robert-Hooke-Straße 5
28359 Bremen, Germany

Tel.: +49 421 178 45-4021
Empfang: +49 421 178 45-6600
Fax: +49 421 178 45-4150
E-Mail: chtz@xxxxxxxxxxxxxxxxxxxxxxxx

Weitere Informationen: http://www.informatik.uni-bremen.de/robotik

--
Benoit