Library issue #1

IainNZ · 2013-10-10T01:50:45Z

idunning@IAINLAPTOP:~/.../Ipopt/test$ julia hs071.jl 
Ptr{Void} @0x00000000047f1e70
Exception of type: OPTION_INVALID in file "IpAlgBuilder.cpp" at line 321:
 Exception message: Selected linear solver MA27 not available.
Tried to obtain MA27 from shared library "libhsl.so", but the following error occured:
libhsl.so: cannot open shared object file: No such file or directory

EXIT: Invalid option encountered.

The text was updated successfully, but these errors were encountered:

IainNZ · 2013-10-10T01:50:59Z

@mlubin

mlubin · 2013-10-10T02:36:42Z

I guess we needed a sparse linear algebra library. It should work now if you wipe deps/usr and deps/src and rebuild.

IainNZ · 2013-10-10T02:47:35Z

Cool, now it just segfaults - progress!

joehuchette · 2014-03-11T23:41:35Z

@IainNZ how did you end up fixing this? I'm havingthe same problem. What deps folder did you wipe?

IainNZ · 2014-03-12T00:13:17Z

I guess it was the deps/usr folder you get after Pkg.build

mlubin · 2014-03-12T00:25:30Z

@joehuchette: you'll definitely need gfortran installed to compile Ipopt with its dependencies

joehuchette · 2014-03-12T00:40:34Z

Yeah I do, looks like it's an issue with the Homebrew version of Ipopt, not with Julia

tkelman · 2014-03-13T16:18:53Z

The default setup from brew install ipopt (at least from homebrew-science, you guys might have a separate version in juliadeps in which case ignore this) builds Mumps using MPI, which you usually don't want to do with the normal release version of Ipopt. Try brew install ipopt --without-mpi, if you're getting it from homebrew-science.

Okay now I see staticfloat/homebrew-juliadeps#9 - you'll want to bump Ipopt to 3.11.7 like Miles just did for Linux, there were bug fixes for Mavericks due to the clang C++ standard library change that aren't in 3.11.5.

joehuchette · 2014-03-13T16:39:28Z

That's good to know, but something else seems to be causing a segfault. Looks like an OpenBLAS issue

tkelman · 2014-03-13T16:52:05Z

Can you run Ipopt's test examples outside of Julia? If it's only segfaulting when running from Julia, then this is probably related to issues that I've debugged to death with the Matlab interface to Ipopt. Matlab (and presumably Julia too when you're using OpenBlas) uses ILP64 BLAS on 64-bit machines. Ipopt and most system BLAS libraries are LP64. If Julia's BLAS and the Ipopt library are both loaded into the same address space, there can be some cross-talk and bad things can happen when integers are the wrong size. Can you post some of what the segfault's outputting, as much of a backtrace as you can get?

If this is what's happening, then it might be easier to statically link the reference BLAS (or some other static-lib LP64 BLAS, ATLAS might work too?) into the Ipopt shared library and live with the slight performance loss compared to OpenBlas. That's what I did with the Ipopt Matlab interface, and for sparse problems the choice of BLAS isn't that big a deal. The multifrontal block reductions in Mumps/MA57 are rarely big enough for OpenBlas to outperform Netlib by enough of a margin to be worth the hassle of messing with Julia/OpenBlas compilation options likeUSE_BLAS64.

joehuchette · 2014-03-13T17:07:39Z

make test works fine outside Julia, so you may be on to something. I get

>> lldb -- julia runtests.jl 
Current executable set to 'julia' (x86_64).
(lldb) r
Process 50391 launched: '/usr/local/julia/julia' (x86_64)

******************************************************************************
This program contains Ipopt, a library for large-scale nonlinear optimization.
 Ipopt is released as open source code under the Eclipse Public License (EPL).
         For more information visit http://projects.coin-or.org/Ipopt
******************************************************************************

NOTE: You are using Ipopt by default with the MUMPS linear solver.
      Other linear solvers might be more efficient (see Ipopt documentation).


This is Ipopt version 3.11.4, running with linear solver mumps.

Number of nonzeros in equality constraint Jacobian...:        4
Number of nonzeros in inequality constraint Jacobian.:        4
Number of nonzeros in Lagrangian Hessian.............:       10

Process 50391 stopped
* thread #1: tid = 0x3791bd, 0x0000000105a60617 libopenblas.dylib`.L41 + 7, queue = 'com.apple.main-thread', stop reason = signal SIGSEGV
    frame #0: 0x0000000105a60617 libopenblas.dylib`.L41 + 7
libopenblas.dylib`.L41 + 7:
-> 0x105a60617:  movhps (%rsi), %xmm0
   0x105a6061a:  addq   %rdx, %rsi
   0x105a6061d:  movsd  (%rsi), %xmm1
   0x105a60621:  addq   %rdx, %rsi
(lldb) bt
* thread #1: tid = 0x3791bd, 0x0000000105a60617 libopenblas.dylib`.L41 + 7, queue = 'com.apple.main-thread', stop reason = signal SIGSEGV
  * frame #0: 0x0000000105a60617 libopenblas.dylib`.L41 + 7
    frame #1: 0x000000010abf8de2 libipopt.dylib`Ipopt::IpBlasDcopy(int, double const*, int, double*, int) + 82
    frame #2: 0x000000010aa2e32b libipopt.dylib`Ipopt::TNLPAdapter::GetStartingPoint(this=0x0000000101681b60, x=<unavailable>, need_x=true, y_c=<unavailable>, need_y_c=false, y_d=<unavailable>, need_y_d=false, z_L=<unavailable>, need_z_L=false, z_U=<unavailable>, need_z_U=false) + 1035 at IpTNLPAdapter.cpp:1488
    frame #3: 0x000000010aab25f9 libipopt.dylib`Ipopt::GradientScaling::DetermineScalingParametersImpl(Ipopt::SmartPtr<Ipopt::VectorSpace const>, Ipopt::SmartPtr<Ipopt::VectorSpace const>, Ipopt::SmartPtr<Ipopt::VectorSpace const>, Ipopt::SmartPtr<Ipopt::MatrixSpace const>, Ipopt::SmartPtr<Ipopt::MatrixSpace const>, Ipopt::SmartPtr<Ipopt::SymMatrixSpace const>, Ipopt::Matrix const&, Ipopt::Vector const&, Ipopt::Matrix const&, Ipopt::Vector const&, double&, Ipopt::SmartPtr<Ipopt::Vector>&, Ipopt::SmartPtr<Ipopt::Vector>&, Ipopt::SmartPtr<Ipopt::Vector>&) + 473
    frame #4: 0x000000010ab18849 libipopt.dylib`Ipopt::StandardScalingBase::DetermineScaling(Ipopt::SmartPtr<Ipopt::VectorSpace const>, Ipopt::SmartPtr<Ipopt::VectorSpace const>, Ipopt::SmartPtr<Ipopt::VectorSpace const>, Ipopt::SmartPtr<Ipopt::MatrixSpace const>, Ipopt::SmartPtr<Ipopt::MatrixSpace const>, Ipopt::SmartPtr<Ipopt::SymMatrixSpace const>, Ipopt::SmartPtr<Ipopt::MatrixSpace const>&, Ipopt::SmartPtr<Ipopt::MatrixSpace const>&, Ipopt::SmartPtr<Ipopt::SymMatrixSpace const>&, Ipopt::Matrix const&, Ipopt::Vector const&, Ipopt::Matrix const&, Ipopt::Vector const&) + 521
    frame #5: 0x000000010ab27ef8 libipopt.dylib`Ipopt::OrigIpoptNLP::InitializeStructures(Ipopt::SmartPtr<Ipopt::Vector>&, bool, Ipopt::SmartPtr<Ipopt::Vector>&, bool, Ipopt::SmartPtr<Ipopt::Vector>&, bool, Ipopt::SmartPtr<Ipopt::Vector>&, bool, Ipopt::SmartPtr<Ipopt::Vector>&, bool, Ipopt::SmartPtr<Ipopt::Vector>&, Ipopt::SmartPtr<Ipopt::Vector>&) + 4264
    frame #6: 0x000000010aaf4400 libipopt.dylib`Ipopt::IpoptData::InitializeDataStructures(Ipopt::IpoptNLP&, bool, bool, bool, bool, bool) + 352
    frame #7: 0x000000010aa94e99 libipopt.dylib`Ipopt::DefaultIterateInitializer::SetInitialIterates() + 185
    frame #8: 0x000000010aabcf9c libipopt.dylib`Ipopt::IpoptAlgorithm::InitializeIterates() + 60
    frame #9: 0x000000010aabb3a5 libipopt.dylib`Ipopt::IpoptAlgorithm::Optimize(bool) + 613
    frame #10: 0x000000010aa01d40 libipopt.dylib`Ipopt::IpoptApplication::call_optimize(this=0x000000010af047c0) + 4464 at IpIpoptApplication.cpp:857
    frame #11: 0x000000010aa0046d libipopt.dylib`Ipopt::IpoptApplication::OptimizeNLP(this=0x000000010af047c0, nlp=0x000000010af04810, alg_builder=0x00007fff5fbfdc98) + 1453 at IpIpoptApplication.cpp:751
    frame #12: 0x000000010a9ffe6b libipopt.dylib`Ipopt::IpoptApplication::OptimizeNLP(this=0x000000010af047c0, nlp=0x000000010af04810) + 91 at IpIpoptApplication.cpp:714
    frame #13: 0x000000010a9ff6c4 libipopt.dylib`Ipopt::IpoptApplication::OptimizeTNLP(this=0x000000010af047c0, tnlp=0x00007fff5fbfdf78) + 228 at IpIpoptApplication.cpp:693
    frame #14: 0x000000010aa09ef7 libipopt.dylib`IpoptSolve(ipopt_problem=0x000000010af05370, x=0x000000010c8c62d0, g=0x000000010b208fb0, obj_val=0x0000000102572ff0, mult_g=0x0000000000000000, mult_x_L=0x0000000000000000, mult_x_U=0x0000000000000000, user_data=0x00000001024bb500) + 1703 at IpStdCInterface.cpp:270
    frame #15: 0x000000010107d723
    frame #16: 0x000000010107d5f3
    frame #17: 0x000000010005b075 libjulia.dylib`jl_apply_generic + 229
    frame #18: 0x00000001000993c1 libjulia.dylib`do_call + 209
    frame #19: 0x0000000100097e96 libjulia.dylib`eval + 694
    frame #20: 0x00000001000980ed libjulia.dylib`eval + 1293
    frame #21: 0x00000001000990fb libjulia.dylib`eval_body + 635
    frame #22: 0x00000001000992a2 libjulia.dylib`jl_interpret_toplevel_thunk_with + 274
    frame #23: 0x00000001000a5e43 libjulia.dylib`jl_toplevel_eval_flex + 1347
    frame #24: 0x00000001000a63a4 libjulia.dylib`jl_parse_eval_all + 276
    frame #25: 0x00000001000a658f libjulia.dylib`jl_load + 127
    frame #26: 0x000000010336ec6b sys.dylib`julia_include1851 - 18446744069360653204
    frame #27: 0x00000001034b62fb sys.dylib`julia_include_from_node115026 - 18446744069359312132
    frame #28: 0x000000010005b075 libjulia.dylib`jl_apply_generic + 229
    frame #29: 0x00000001000993c1 libjulia.dylib`do_call + 209
    frame #30: 0x0000000100097e96 libjulia.dylib`eval + 694
    frame #31: 0x00000001000a5dd2 libjulia.dylib`jl_toplevel_eval_flex + 1234
    frame #32: 0x00000001000a63a4 libjulia.dylib`jl_parse_eval_all + 276
    frame #33: 0x00000001000a658f libjulia.dylib`jl_load + 127
    frame #34: 0x000000010336ec6b sys.dylib`julia_include1851 - 18446744069360653204
    frame #35: 0x0000000101069632
    frame #36: 0x000000010005b075 libjulia.dylib`jl_apply_generic + 229

mlubin · 2014-03-13T17:12:44Z

@tkelman: that's a good point. Perhaps we can make a configure flag for Ipopt to use ILP64 for blas/lapack.

tkelman · 2014-03-13T17:25:37Z

@joehuchette Yep, exactly what I thought it was. See https://projects.coin-or.org/Ipopt/browser/releases/3.11.4/Ipopt/src/LinAlg/IpBlas.cpp#L111 for definition of IpBlasDcopy. Ipopt on a Mac will use framework vecLib for Blas and Lapack if you don't tell it otherwise, I'm guessing you didn't specifically ask Ipopt to use OpenBlas here?

@mlubin Yes we should look into this, but I don't think it will be an easy change. It would take quite a bit of digging to find everywhere that ipfint is being used interchangeably with int or Index without typecasting. There's a FIXME about it here: https://projects.coin-or.org/Ipopt/browser/releases/3.11.4/Ipopt/configure.ac#L433, but currently it's just hardcoded to int - https://projects.coin-or.org/Ipopt/browser/releases/3.11.4/Ipopt/src/Common/IpTypes.hpp#L26. All the sparse solvers and user interfaces are also currently using int, it could get messy very quickly.

I don't think anyone has tried hacking at this one, at least not AFAIK. The easiest solution is statically linking BLAS and LAPACK into the Ipopt shared library, but for obvious reasons that's not good enough to be a permanent fix.

mlubin · 2014-03-14T22:13:30Z

With @joehuchette here. Ipopt depends on vecLib, not openblas, so I'm not sure why openblas is being called. How do you go about statically linking the reference blas?

tkelman · 2014-03-14T23:03:50Z

I'm not sure why openblas is being called

Name shadowing on dcopy_ etc.

How do you go about statically linking the reference blas?

Easiest way is configuring Ipopt --with-blas=BUILD --with-lapack=BUILD --disable-shared but then you won't get a shared Ipopt library which Julia needs.

To get a shared Ipopt library statically linked to reference blas, try something like

for i in Blas Lapack; do
  cd ThirdParty/$i
  ./get.$i
  ./configure --disable-shared --with-pic --prefix=$prefix
  make install
  cd ../..
done
./configure --with-blas="-L$prefix/lib -lcoinblas" \
  --with-lapack="-L$prefix/lib -lcoinlapack" \
  --prefix=$prefix --enable-dependency-linking
make install

tkelman · 2014-03-14T23:13:46Z

Should add a test that uses hessian_approximation = limited-memory for good measure, as that will call Lapack.

joehuchette · 2014-03-14T23:58:04Z

Hmm I tried this, and it still chokes on an openblas call.

tkelman · 2014-03-14T23:59:56Z

What do ldd libipopt.so or otool -L libipopt.dylib say?

joehuchette · 2014-03-15T00:01:15Z

>> otool -L libipopt.dylib 
libipopt.dylib:
    /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib/libipopt.1.dylib (compatibility version 11.0.0, current version 11.4.0)
    /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib/libcoinmumps.1.dylib (compatibility version 6.0.0, current version 6.9.0)
    /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1197.1.1)
    /usr/local/Cellar/gfortran/4.8.2/gfortran/lib/libgfortran.3.dylib (compatibility version 4.0.0, current version 4.0.0)
    /usr/local/Cellar/gfortran/4.8.2/gfortran/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0)
    /usr/local/Cellar/gfortran/4.8.2/gfortran/lib/libquadmath.0.dylib (compatibility version 1.0.0, current version 1.0.0)
    /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib/libcoinlapack.1.dylib (compatibility version 6.0.0, current version 6.6.0)
    /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib/libcoinblas.1.dylib (compatibility version 5.0.0, current version 5.6.0)
    /usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 120.0.0

tkelman · 2014-03-15T00:03:36Z

Can you clear out /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib, re-run the make for ThirdParty/Blas and ThirdParty/Lapack outputting to a log file, and gist the logs, along with ThirdParty/Blas/config.log and ThirdParty/Lapack/config.log?

joehuchette · 2014-03-15T00:11:53Z

Here you go:
https://gist.github.com/joehuchette/2e5d7ea0bcc0a450ea70

Thanks for helping to debug, BTW.

tkelman · 2014-03-15T00:35:29Z

It's reusing libtool from the (previous) Ipopt configure when building Blas and Lapack, so not obeying --disable-shared properly. First, delete libtool from the top-level Ipopt directory. Get rid of as much as you can from past builds, preferably start from a clean source tree (plus ./get.Mumps). I try to avoid in-source builds for this very reason.

Just to be sure, let's use a separate build folder, and also be a little more explicit about wanting static libs:

for i in Blas Lapack; do
  cd ThirdParty/$i
  ./get.$i
  mkdir -p build
  cd build
  ../configure --disable-shared --enable-static --with-pic --prefix=$prefix
  make install
  cd ../../..
done
mkdir -p build
cd build
../configure --with-blas=$prefix/lib/libcoinblas.a \
  --with-lapack=$prefix/lib/libcoinlapack.a \
  --prefix=$prefix --enable-dependency-linking
make install

Then let me know what otool -L libipopt.dylib says again.
And ls /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib

joehuchette · 2014-03-15T04:27:59Z

I run

prefix=/Users/huchette/.julia/v0.3/Ipopt/deps/usr

cd ThirdParty/Mumps
./get.Mumps
mkdir -p build
cd build
../configure --prefix=$prefix --disable-shared --enable-static --without-lapack --without-mpi
make install
cd ../..

for i in Blas Lapack; do
  cd $i
  ./get.$i
  mkdir -p build
  cd build
  ../configure --disable-shared --enable-static --with-pic --without-mpi --prefix=$prefix
  make install
  cd ../..
done
cd ..
mkdir -p build
cd build
../configure --with-blas=$prefix/lib/libcoinblas.a \
  --with-lapack=$prefix/lib/libcoinlapack.a \
  --prefix=$prefix --enable-dependency-linking --without-mpi
make install

and get

>> otool -L libipopt.dylib
libipopt.dylib:
    /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib/libipopt.1.dylib (compatibility version 11.0.0, current version 11.4.0)
    /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib/libcoinmumps.1.dylib (compatibility version 6.0.0, current version 6.9.0)
    /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1197.1.1)
    /usr/local/Cellar/gfortran/4.8.2/gfortran/lib/libgfortran.3.dylib (compatibility version 4.0.0, current version 4.0.0)
    /usr/local/Cellar/gfortran/4.8.2/gfortran/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0)
    /usr/local/Cellar/gfortran/4.8.2/gfortran/lib/libquadmath.0.dylib (compatibility version 1.0.0, current version 1.0.0)
    /usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 120.0.0)
>> ls -l /Users/huchette/.julia/v0.3/Ipopt/deps/usr/lib
total 14152
-rw-r--r--  1 huchette  staff   196512 Mar 15 00:18 libcoinblas.a
-rwxr-xr-x  1 huchette  staff      784 Mar 15 00:18 libcoinblas.la
-rw-r--r--  1 huchette  staff   737872 Mar 15 00:19 libcoinlapack.a
-rwxr-xr-x  1 huchette  staff      790 Mar 15 00:19 libcoinlapack.la
-rwxr-xr-x  1 huchette  staff  1714352 Mar 15 00:21 libcoinmumps.1.4.9.dylib
lrwxr-xr-x  1 huchette  staff       24 Mar 15 00:21 libcoinmumps.1.dylib -> libcoinmumps.1.4.9.dylib
-rw-r--r--  1 huchette  staff  1964160 Mar 15 00:17 libcoinmumps.a
lrwxr-xr-x  1 huchette  staff       24 Mar 15 00:21 libcoinmumps.dylib -> libcoinmumps.1.4.9.dylib
-rwxr-xr-x  1 huchette  staff      857 Mar 15 00:21 libcoinmumps.la
-rwxr-xr-x  1 huchette  staff  2589152 Mar 15 00:23 libipopt.1.9.4.dylib
lrwxr-xr-x  1 huchette  staff       20 Mar 15 00:23 libipopt.1.dylib -> libipopt.1.9.4.dylib
lrwxr-xr-x  1 huchette  staff       20 Mar 15 00:23 libipopt.dylib -> libipopt.1.9.4.dylib
-rwxr-xr-x  1 huchette  staff     1300 Mar 15 00:23 libipopt.la
drwxr-xr-x  6 huchette  staff      204 Mar 15 00:23 pkgconfig

tkelman · 2014-03-15T04:35:55Z

I think we're in business then, does that a) succeed at Ipopt's make test, and b) work in Julia?

You shouldn't need to worry about --without-mpi anywhere, that's something that should only matter if you're getting Ipopt from homebrew-science. Or maybe if you have some MPICC etc. environment variables set. Was it causing problems without that?

You also shouldn't need to configure or build Mumps separately, just download its source code with ./get.Mumps before you configure Ipopt. A dylib for Mumps should be fine (and was what you got, since you didn't tell Ipopt anything about the separate static version you built without needing to).

joehuchette · 2014-03-15T05:12:05Z

The unit tests are choking on a missing symbol in Mumps:

./run_unitTests

Running unitTests...

Testing AMPL Solver Executable...
    no AMPL solver executable found, skipping test...
Testing C++ Example...
./run_unitTests: line 59: 58087 Trace/BPT trap: 5       ./hs071_cpp > tmpfile 2>&1

 ---- 8< ---- Start of test program output ---- 8< ----
dyld: lazy symbol binding failed: Symbol not found: _dsyr_
  Referenced from: /Users/huchette/.julia/v0.3/Ipopt/deps/src/Ipopt-3.11.4/build/ThirdParty/Mumps/.libs/libcoinmumps.1.dylib
  Expected in: flat namespace

dyld: Symbol not found: _dsyr_
  Referenced from: /Users/huchette/.julia/v0.3/Ipopt/deps/src/Ipopt-3.11.4/build/ThirdParty/Mumps/.libs/libcoinmumps.1.dylib
  Expected in: flat namespace

 ---- 8< ----  End of test program output  ---- 8< ----

    ******** Test FAILED! ********
Output of the test program is above.
Testing C Example...
./run_unitTests: line 77: 58105 Trace/BPT trap: 5       ./hs071_c > tmpfile 2>&1

 ---- 8< ---- Start of test program output ---- 8< ----
dyld: lazy symbol binding failed: Symbol not found: _dsyr_
  Referenced from: /Users/huchette/.julia/v0.3/Ipopt/deps/src/Ipopt-3.11.4/build/ThirdParty/Mumps/.libs/libcoinmumps.1.dylib
  Expected in: flat namespace

dyld: Symbol not found: _dsyr_
  Referenced from: /Users/huchette/.julia/v0.3/Ipopt/deps/src/Ipopt-3.11.4/build/ThirdParty/Mumps/.libs/libcoinmumps.1.dylib
  Expected in: flat namespace

 ---- 8< ----  End of test program output  ---- 8< ----

    ******** Test FAILED! ********
Output of the test program is above.
Testing Fortran Example...
./run_unitTests: line 95: 58122 Trace/BPT trap: 5       ./hs071_f > tmpfile 2>&1

 ---- 8< ---- Start of test program output ---- 8< ----
dyld: lazy symbol binding failed: Symbol not found: _dsyr_
  Referenced from: /Users/huchette/.julia/v0.3/Ipopt/deps/src/Ipopt-3.11.4/build/ThirdParty/Mumps/.libs/libcoinmumps.1.dylib
  Expected in: flat namespace

dyld: Symbol not found: _dsyr_
  Referenced from: /Users/huchette/.julia/v0.3/Ipopt/deps/src/Ipopt-3.11.4/build/ThirdParty/Mumps/.libs/libcoinmumps.1.dylib
  Expected in: flat namespace


Program received signal SIGTRAP: Trace/breakpoint trap.

Backtrace for this error:
#0  0x10a069f72
#1  0x10a06a73e
#2  0x7fff895bf5a9
 ---- 8< ----  End of test program output  ---- 8< ----

    ******** Test FAILED! ********
Output of the test program is above.
make[2]: *** [test] Error 255
make[1]: *** [unitTest] Error 2
make: *** [test] Error 2

tkelman · 2014-03-15T06:28:05Z

Damn, the unit tests ran in Linux but apparently not on Mac. I'll fire up my Mac VM and let you know what I figure out. Might involve building Mumps statically too, like you did but telling Ipopt about it a la --with-mumps=$prefix/lib/libcoinmumps.a

tkelman · 2014-03-15T08:25:53Z

Can reproduce, more or less, with Ipopt 3.11.4.

Please update to Ipopt 3.11.7.

Used this build script, essentially identical to this #1 (comment)

joehuchette · 2014-03-15T15:54:08Z

Wow, looks like this works! Thanks so much!

mlubin · 2014-03-15T15:54:44Z

Nice! Just need to update the homebrew script now.

tkelman · 2014-03-15T15:56:29Z

I have a fixed version of the BinDeps SimpleBuild, but didn't do the homebrew script.

@warn

* @warn instead of warn(

Update

mlubin closed this as completed in 0c35003 Oct 10, 2013

joehuchette mentioned this issue Mar 10, 2014

Build error on OSX #3

Closed

This was referenced Mar 16, 2014

bump Ipopt staticfloat/homebrew-juliadeps#10

Merged

ABI conflicts due to 64-bit libopenblas.so JuliaLang/julia#4923

Closed

tkelman mentioned this issue Apr 10, 2014

Using Homebrew MUMPS JuliaInv/MUMPSjInv.jl#2

Open

tkelman mentioned this issue Sep 29, 2014

pre-build libraries for Julia packages and create default git config JuliaCloud/JuliaBox#103

Merged

donboyd5 mentioned this issue May 28, 2017

Wondering if I can use pardiso Windows dll as linear solver for Ipopt without getting recompiled Ipopt #83

Closed

mlubin pushed a commit that referenced this issue Oct 6, 2018

@warn instead of warn( (#1) (#145)

f9201dd

* @warn instead of warn(

blegat pushed a commit that referenced this issue Jun 10, 2020

Merge pull request #1 from JuliaOpt/master

e989904

Update

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Library issue #1

Library issue #1

IainNZ commented Oct 10, 2013

IainNZ commented Oct 10, 2013

mlubin commented Oct 10, 2013

IainNZ commented Oct 10, 2013

joehuchette commented Mar 11, 2014

IainNZ commented Mar 12, 2014

mlubin commented Mar 12, 2014

joehuchette commented Mar 12, 2014

tkelman commented Mar 13, 2014

joehuchette commented Mar 13, 2014

tkelman commented Mar 13, 2014

joehuchette commented Mar 13, 2014

mlubin commented Mar 13, 2014

tkelman commented Mar 13, 2014

mlubin commented Mar 14, 2014

tkelman commented Mar 14, 2014

tkelman commented Mar 14, 2014

joehuchette commented Mar 14, 2014

tkelman commented Mar 14, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

mlubin commented Mar 15, 2014

tkelman commented Mar 15, 2014

Library issue #1

Library issue #1

Comments

IainNZ commented Oct 10, 2013

IainNZ commented Oct 10, 2013

mlubin commented Oct 10, 2013

IainNZ commented Oct 10, 2013

joehuchette commented Mar 11, 2014

IainNZ commented Mar 12, 2014

mlubin commented Mar 12, 2014

joehuchette commented Mar 12, 2014

tkelman commented Mar 13, 2014

joehuchette commented Mar 13, 2014

tkelman commented Mar 13, 2014

joehuchette commented Mar 13, 2014

mlubin commented Mar 13, 2014

tkelman commented Mar 13, 2014

mlubin commented Mar 14, 2014

tkelman commented Mar 14, 2014

tkelman commented Mar 14, 2014

joehuchette commented Mar 14, 2014

tkelman commented Mar 14, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

tkelman commented Mar 15, 2014

tkelman commented Mar 15, 2014

joehuchette commented Mar 15, 2014

mlubin commented Mar 15, 2014

tkelman commented Mar 15, 2014