First login?
Lost password?
 
 
cellml-tools-developers AT lists.cellml.org 
A list for the developers of CellML tools

Text archives Help


RE: [[cellml-dev] ] Mac OS X CellML API 1.12


Chronological Thread 
  • From: "Alan Garny" <alan.garny AT inria.fr>
  • To: <cellml-tools-developers AT lists.cellml.org>
  • Subject: RE: [[cellml-dev] ] Mac OS X CellML API 1.12
  • Date: Wed, 14 May 2014 09:52:33 +0200
  • Organization: INRIA
Hi Ivo,



Just to confirm that OpenCOR does indeed use RPATH, but only on Linux (what
is certain is that you really don’t want to use DYLD_LIBRARY_PATH; its use is
really not ‘cool’).



On OS X, I do what is considered the correct answer in the link you gave
(i.e.
http://stackoverflow.com/questions/10021428/macos-how-to-link-a-dynamic-library-with-a-relative-path-using-gcc-ld).
If you use CMake, here is how I do it: look for the various instances of
“install_name_tool” in
https://github.com/opencor/opencor/blob/master/cmake/common.cmake. Regarding
David’s CMake link, I wasn’t aware of CMake having a bundle utilities module,
so no idea about that one.



Alan



From:
cellml-tools-developers-request AT lists.cellml.org

[mailto:cellml-tools-developers-request AT lists.cellml.org]
On Behalf Of David Nickerson
Sent: 14 May 2014 02:07
To:
cellml-tools-developers AT lists.cellml.org
Subject: Re: [[cellml-dev] ] Mac OS X CellML API 1.12



Hi Ivo,



You can also make use of the DYLD_LIBRARY_PATH to point your environment at
the appropriate location to find the libraries at run time.



In my tools, I use CMake to help "install" my applications and part of this
process will collect libraries and put them in the appropriate location for
the application and rewrite things like the rpath to be the correct values.
This also makes it trivial to then create archives for distribution. I
believe Alan Garny does the same thing in OpenCOR. See
http://www.cmake.org/cmake/help/v3.0/module/BundleUtilities.html for some
more hints on how this is done with CMake.



As OpenCMISS(Iron) and the OpenCMISSExtras move over to using CMake, build
support for platforms other than Linux will become available "natively".





Cheers,

David.



On Wed, May 14, 2014 at 11:54 AM, Ivo Siekmann
<ivo.siekmann AT unimelb.edu.au>
wrote:

Hi David (and others interested in the Mac OS X binaries),

I was able to link to the dynamic libraries provided in the Cell ML SDK
package... but it is not very convenient to use them.

I believe that the "install name" in the .dylib files (see, for example, the
response to this question)
http://stackoverflow.com/questions/10021428/macos-how-to-link-a-dynamic-library-with-a-relative-path-using-gcc-ld
is set to "./". Therefore linking to the libraries in the Cell ML SDK works
but running the executable fails unless all .dylib files are copied to the
directory of the executable. I'm not sure if this is the reason that the Cell
ML SDK package is shown as "damaged or incomplete" by Mac OS X 10.9. And I
don't know enough about building dynamic libraries on Mac OS X that I would
have an idea how to fix this. But I'll be happy to enter this in the tracker
if you think this is a good idea?!?

Cheers
Ivo

P.S.: I managed to compile the API from source and link to OpenCMISS (which
was what I wanted), so thanks for the hint!



On 6/05/14 10:29 am, David Nickerson wrote:

Hi Ivo,

See https://tracker.physiomeproject.org/show_bug.cgi?id=3673,
https://tracker.physiomeproject.org/show_bug.cgi?id=3671, and
https://tracker.physiomeproject.org/show_bug.cgi?id=3670 for some
recent tracker items regarding building the CellML API under OS X
10.9.

The upshot is that you should be able to build the current trunk code
(which has some fixes that are not in the 1.12 release, so I would
recommend using that), but you will not be able to build all the
tests. I have been able to build CellML2C (the code generation test)
but haven't tried the other test codes. I use clang for all
compilation/linking on OS X:
$ clang --version
Apple LLVM version 5.1 (clang-503.0.40) (based on LLVM 3.4svn)
Target: x86_64-apple-darwin13.1.0
Thread model: posix

The 1.12 binaries should work - not sure why you're getting that
error. I know I have linked my own code agains the 1.12 binaries
(using clang), but generally use my own build of the current trunk
code nowadays.

If needed, Alan Garny has the OpenCOR development environment all
nicely set up so you can probably extract the required binaries from
that source.


Cheers,
Andre.

On Mon, May 5, 2014 at 5:12 PM, Ivo Siekmann
<ivo.siekmann AT unimelb.edu.au>
wrote:

Hi CellMLers,

I'm trying to get the CellML API running on Mac and I ran into some
difficulties.

1) Compiling from source code:
a) This fails with clang (Apple LLVM version 5.1 (clang-503.0.40)),
apparently due to a forward declaration of the << operator in TestAssert.h
when attempting to build DOMTEST.cpp.o (I don't know much about this but
from some googling it seems to me that clang is stricter about the standard
than some other compilers. So it may see errors where other compilers
don't).

b) Compiling with gcc 4.8.2 goes through and passes the tests of
'CheckCodeGenerator' and 'CheckVACSS' but fails for 'RunTestBin'. I didn't
find out a lot more about how/why it failed.

2) Binaries:
I've downloaded the binaries and Mac OS X complains that the package can't
be opened "because it may be damaged or incomplete". Well, I can open the
package and look at its contents, so it may be possible to link the
libraries but I haven't tried.

Any help would be greatly appreciated.
Cheers
Ivo









--





David Nickerson on about.me
<http://d13pix9kaak6wt.cloudfront.net/signature/me-badge.png>



David Nickerson

about.me/david.nickerson







Archive powered by MHonArc 2.6.18.

Top of page