[Ipopt] Matlab Build with -framework Accelerate

[Ted Ralphs]

I made a new BuildTools release just a few days ago that should address
this issue. Here is the changeset:

https://projects.coin-or.org/BuildTools/changeset/3273/

For this change to propagate, though, we would need to make new releases of
all the dependent projects. I would guess that re-running autoconf on all
dependencies to produce new configure scripts would  fix the issue. We'll
try to do this as quickly as possible. I believe a new stable version of
Ipopt is also in the works and that should be available within about 2
weeks, but Stefan can confirm this.

If you need this to be fixed quickly, I can tell you what to try yourself
in more detail.

Cheers,

Ted

On Tue, Jan 13, 2015 at 8:14 AM, Marco Rozgic <marco.rozgic at hsu-hh.de>
wrote:

> Hi everybody,
>
> I am currently building Ipopt 3.11.9 on MacOS 10.10 with Matlab2014b. I
> managed to get a working mex file with some effort that I am willing to
> share. However, afterwards I was trying to build a mex file that
> incorporates the MacOS blas and lapack, wich are now given by the framework
> Accelerate and not anymore by the vecLib framework. I used the configure
> site script and added the following to the configure command:
>
> ../configure --enable-matlab-ma57 --with-blas='-framework Accelerate'
> --with-lapack='-framework Accelerate‘
>
> Afterwards I run
>
> make
> make test
> make install
>
> All good up to now. I create a mexopt.sh file, edit the SDKs to the newest
> versions and modify the makefile for the Matlab interface. Now make gives
> the following error:
>
> Undefined symbols for architecture x86_64:
>
> Followed by some lapack and blas routines that are not found, e.g.,
>
>  "_dgemm_", referenced from:
>       Ipopt::IpBlasDgemm(bool, bool, int, int, int, double, double const*,
> int, double const*, int, double, double*, int) in libipopt.a(IpBlas.o)
>       _ma57td_ in libcoinhsl.a(ma57d.o)
>       _ma57rd_ in libcoinhsl.a(ma57d.o)
>       _ma57qd_ in libcoinhsl.a(ma57d.o)
>       _ma57od_ in libcoinhsl.a(ma57d.o)
>
> So I figure that the Acceleration framework is not passed correctly or
> included correctly.
> Does anybody have an idea how to work around that issue or resolve the
> problem?
>
> Best,
>
> Marco
>
> --
> Dipl.-Math. Marco Rozgic
> Department of the Theory of Electrical Engineering at the
> Helmut Schmidt University
> University of the Federal Armed Forces Hamburg
> Tel: +49 (0) 40 6541 3647
> Location: Holstenhofweg 85, 22043 Hamburg Building H3 Room 007
>
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>
>


-- 
Dr. Ted Ralphs
Professor, Lehigh University
(610) 628-1280
ted 'at' lehigh 'dot' edu
coral.ie.lehigh.edu/~ted
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20150113/6729e11a/attachment.html>