New Ticket     Tickets     Wiki     Browse Source     Timeline     Roadmap     Ticket Reports     Search

Ticket #31833 (closed defect: fixed)

Opened 2 years ago

Last modified 16 months ago

py27-scipy fails to build with -faltivec flag error

Reported by: martinzinkin@… Owned by: ram@…
Priority: Normal Milestone:
Component: ports Version: 2.0.3
Keywords: Cc: calsaiwing@…
Port: py27-scipy

Description

py27-scipy fails to build on my Mac with the errors below - apparently complaining about an unsupported -faltivec compiler flag

I am on OS X Lion 10.7.2, Xcode 4.2 (both fully up to date), and I have just run selfupdate and have no outdated ports.

I have tried both gcc44 and gcc variants of py27-scipy with the same results.

this old numpy issue looks similar, but I am afraid I am not sufficiently familiar with macports to make further progress.

:info:build compiling C sources
:info:build C compiler: /opt/local/bin/gcc-mp-4.4 -fno-strict-aliasing -fno-common -dynamic -pipe -O2 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes
:info:build 
:info:build compile options: '-DNO_ATLAS_INFO=3 -I/Library/Python/2.7/site-packages/numpy-2.0.0.dev_e2af7b7_20110721-py2.7-macosx-10.7-x86_64.egg/numpy/core/include -I/opt/local/Library/Frameworks/Python.framework/Versions/2.7/include/python2.7 -c'
:info:build extra options: '-faltivec -I/System/Library/Frameworks/vecLib.framework/Headers'
:info:build gcc-mp-4.4: scipy/integrate/_odepackmodule.c
:info:build cc1: error: unrecognized command line option "-faltivec"
:info:build cc1: error: unrecognized command line option "-faltivec"
:info:build error: Command "/opt/local/bin/gcc-mp-4.4 -fno-strict-aliasing -fno-common -dynamic -pipe -O2 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -DNO_ATLAS_INFO=3 -I/Library/Python/2.7/site-packages/numpy-2.0.0.dev_e2af7b7_20110721-py2.7-macosx-10.7-x86_64.egg/numpy/core/include -I/opt/local/Library/Frameworks/Python.framework/Versions/2.7/include/python2.7 -c scipy/integrate/_odepackmodule.c -o build/temp.macosx-10.7-x86_64-2.7/scipy/integrate/_odepackmodule.o -faltivec -I/System/Library/Frameworks/vecLib.framework/Headers" failed with exit status 1

Attachments

log.txt (548.0 KB) - added by martinzinkin@… 2 years ago.
Log file
log.2.txt (548.0 KB) - added by martinzinkin@… 2 years ago.
Log file
log.3.txt (548.0 KB) - added by martinzinkin@… 2 years ago.
Log file
main.log (528.6 KB) - added by samaburden@… 2 years ago.

Change History

Changed 2 years ago by martinzinkin@…

Log file

Changed 2 years ago by martinzinkin@…

Log file

Changed 2 years ago by martinzinkin@…

Log file

comment:1 Changed 2 years ago by macsforever2000@…

  • Keywords p27-scipy lion -faltivec removed
  • Owner changed from macports-tickets@… to ram@…
  • Port changed from p27-scipy to py27-scipy

comment:2 Changed 2 years ago by ryandesign@…

AltiVec is of course a feature of PowerPC G4 and G5 processors; it seem silly that anyone is requesting AltiVec features when compiling on Intel processors.

comment:3 follow-up: ↓ 4 Changed 2 years ago by ram@…

I notice that in the log message there is reference to a numpy-2.0.0 development snapshot install in /Library/Python, this is clearly interfering with the build. I would advise removing this and trying the build again.

comment:4 in reply to: ↑ 3 Changed 2 years ago by martinzinkin@…

Replying to ram@…:

I notice that in the log message there is reference to a numpy-2.0.0 development snapshot install in /Library/Python, this is clearly interfering with the build. I would advise removing this and trying the build again.

Excellent, that was it. Thanks very much.

comment:5 follow-up: ↓ 6 Changed 2 years ago by ram@…

  • Status changed from new to closed
  • Resolution set to invalid

comment:6 in reply to: ↑ 5 ; follow-up: ↓ 15 Changed 2 years ago by samaburden@…

  • Status changed from closed to reopened
  • Resolution invalid deleted

Replying to ram@…:

I have the same build error, but I can't find similar reference to old numpy development snapshot as the original post (see attached log file).

I've tried removing py27-numpy (& deps) and reinstalling, but receive the same error. Thoughts?

Changed 2 years ago by samaburden@…

comment:7 follow-ups: ↓ 8 ↓ 9 Changed 2 years ago by ram@…

Look like it's using stuff from the system numpy:

:info:build building 'scipy.integrate._odepack' extension
:info:build compiling C sources
:info:build C compiler: /opt/local/bin/gcc-mp-4.4 -fno-strict-aliasing -fno-common -dynamic -pipe -O2 -fwrapv -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes
:info:build 
:info:build compile options: '-DNO_ATLAS_INFO=3 -I/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python/numpy/core/include -I/opt/local/Library/Frameworks/Python.framework/Versions/2.7/include/python2.7 -c'
:info:build extra options: '-faltivec -I/System/Library/Frameworks/vecLib.framework/Headers'
:info:build gcc-mp-4.4: scipy/integrate/_odepackmodule.c
:info:build cc1: error: unrecognized command line option "-faltivec"
:info:build cc1: error: unrecognized command line option "-faltivec"

I imagine you're getting this error because the system numpy is being used instead of the MacPorts version. Not sure why this is the case though. I'll do a little digging and see if I can find anything...

comment:8 in reply to: ↑ 7 Changed 2 years ago by samaburden@…

Replying to ram@…:

Look like it's using stuff from the system numpy:

I imagine you're getting this error because the system numpy is being used instead of the MacPorts version. Not sure why this is the case though. I'll do a little digging and see if I can find anything...

Gotcha -- is there a way to prevent the system numpy from being found? I haven't added it to my PYTHONPATH . . .

Thanks!

comment:9 in reply to: ↑ 7 Changed 2 years ago by samaburden@…

Replying to ram@…:

Look like it's using stuff from the system numpy:

You were absolutely right -- I hid /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python and py27-scipy installed successfully.

Thanks for the help!

comment:10 follow-up: ↓ 11 Changed 2 years ago by ram@…

Great, now got to figure out why it was picking up the system numpy and not the MacPorts version...

comment:11 in reply to: ↑ 10 Changed 2 years ago by samaburden@…

Replying to ram@…:

Great, now got to figure out why it was picking up the system numpy and not the MacPorts version...

In case it's relevant: for some reason, the system's Python libraries show up before MacPorts' Python libraries in sys.path. I originally tried installing numpy & scipy without MacPorts, so maybe I tweaked my path along the way?

comment:12 Changed 2 years ago by calsaiwing@…

  • Cc calsaiwing@… added

Cc Me!

comment:13 in reply to: ↑ description Changed 2 years ago by calsaiwing@…

(probably is expected) still happens with MacPorts 2.0.4.

comment:14 Changed 2 years ago by ram@…

This is only expected if you try and link against the system numpy, if you use the MacPorts version then it will build successfully. If you have modified sys.path then I imagine that will cause the problem.

comment:15 in reply to: ↑ 6 Changed 2 years ago by gwoptics@…

Replying to samaburden@…:

Replying to ram@…:

I have the same build error, but I can't find similar reference to old numpy development snapshot as the original post (see attached log file).

I've tried removing py27-numpy (& deps) and reinstalling, but receive the same error. Thoughts?

Another possibility: a numpy version installed previously via easy_install-2.7 causes a `port install py27-scipy' to fail similarly (for me). After:

  • sudo port install py27-pip
  • sudo pip-2.7 uninstall numpy
  • sudo port install py27-scipy

it worked. Maybe that helps.

comment:16 Changed 17 months ago by ram@…

Is this still a problem?

comment:17 Changed 16 months ago by ram@…

  • Status changed from reopened to closed
  • Resolution set to fixed
Note: See TracTickets for help on using tickets.