fail to build atlas in i7 mac

[high07@…]

To set up py26-nltk, I was trying to setup atlas as it is a dependency for nltk.

However, it constantly fails, although I tried most things suggested by related thread. I am not sure what should I do for this. any help will be greatly appreciated.

sudo port clean --all atlas
Password:
--->  Cleaning atlas

 sudo port install atlas
--->  Computing dependencies for atlas
--->  Fetching atlas
--->  Attempting to fetch atlas3.9.37.tar.bz2 from http://voxel.dl.sourceforge.net/math-atlas
--->  Attempting to fetch lapack-3.3.0.tgz from http://www.netlib.org/lapack
--->  Verifying checksum(s) for atlas
--->  Extracting atlas
--->  Applying patches to atlas
--->  Configuring atlas
--->  Building atlas

Error: Target org.macports.build returned: shell command failed (see log for details)
Log for atlas is at: /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_release_ports_math_atlas/main.log
Error: Status 1 encountered during processing.
To report a bug, see <http://guide.macports.org/#project.tickets>

[high07@…]

log file

[Veence (Vincent)]

You should try to compile with gcc45 instead of gcc44. It would not even surprise me if, for these newest model, gcc46 is not required.

[high07@…]

Replying to vince@…:

You should try to compile with gcc45 instead of gcc44. It would not even surprise me if, for these newest model, gcc46 is not required.

Yes somebody suggested the same thing so i tried to do that. I successfully downloaded and built gcc45 but i got those result after setting up gcc45

[Veence (Vincent)]

The log you've posted indicates you've tried to build atlas using gcc44, not gcc45. You should use the +gcc45 option. I'm going to see if it is worth adding a +gcc46 new option. But, as a general rule, ports like Atlas, which rely heavily on the knowledge of the underlying hardware (in this case, CPU), always lag a bit after a new hardware release. You may have to wait several months for an updated version. In this case, except signaling it upstream, there is little we can do.

[Veence (Vincent)]

It seems a 3.9.38 version is out that corrects a bug and is compatible with the new Sandy bridge processors. I will propose an update ASAP.

[Veence (Vincent)]

3.9.38 seems to be buggy, I've signaled it upstream, but it may take a little while to be fixed (assuming this is a real bug and not a configure trick).

[otto.hasselblad@…]

Cc Me!

[Veence (Vincent)]

I have reported some bugs to Atlas' maintainer, and they should be fixed in the new 3.9.41 release I am currently testing, I'll keep you posted.

[j.bugzilla2@…]

Cc Me!

[SlaunchaMan (Jeff Kelley)]

Cc Me!

[jmroot (Joshua Root)]

Does this still happen with 3.9.47?

[Veence (Vincent)]

I assume it works now…