Seconded - we can't possibly fix this without knowing what is actually breaking. -Nat On Mon, Jun 10, 2013 at 2:44 AM, Oleg Dolomanov wrote:

Hi Alastair,

what was the exception in the is_64bit_architecture function? Could you send the stack trace?

Cheers,

Oleg.

Date: Sun, 9 Jun 2013 18:04:09 -0700 From: [email protected] To: [email protected] Subject: [cctbxbb] build problem on ubuntu 12.04

Hello, is the current cctbx build (2013_02_27_0005) known to be buildable on Ubuntu ? I tried "perl cctbx_bundle.selfx" without success. After some poking about I found an exception raised in the platform.architecture() call in env_config.py/is_64bit_architecture(). Modifying the function to simply return False allowed the build to complete (log attached).This is on 12.04 with stock python 2.7.3 and the "platform.architecture()" call works as expected in a simple test. thanks, Alastair

_______________________________________________ cctbxbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/cctbxbb

_______________________________________________ cctbxbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/cctbxbb

On Tue, Jun 11, 2013 at 2:03 PM, Oleg Dolomanov wrote:

it appears that somehow the 'file' command used by Python to determine its architecture fails in that particular setup (note the doc link later) and leads Python to call exit(1) (not bad for safe defaults!). I also discovered that in the documentation (not sure if that was originally there when I was implementing this, http://docs.python.org/2/library/platform.html) says that there are better ways to realise if this is a 32 or 64 bit system (like test: is_64bits = sys.maxsize > 2**32), and I am not sure if we should Python to do that decision (without changing anything) and if the failure actually tells us anything or should use the proposed way?

Okay, I've added the sys.maxsize check if the call to platform.architecture() doesn't work. The only issue with this is that sys.maxsize wasn't introduced until Python 2.6, so we can't use it by default yet. -Nat

Oh, I missed an important fact - platform() MUST! return a best default, if it fails, it causes exit(1), (according to Alastair's stack trace) that cannot be caught anyway (unless you override it beforehand like sys.abort = dummy , sys.exit= dummy), So, I suggest Alastair should fix his platform settings out (I did test 12.04 x32 and x64 bits on a VM-Ware systems and they worked fine). Cheers, o

From: [email protected] Date: Tue, 11 Jun 2013 15:52:08 -0600 To: [email protected] Subject: Re: [cctbxbb] build problem on ubuntu 12.04

On Jun 11, 2013, at 3:41 PM, Nathaniel Echols wrote:

...

Okay, I've added the sys.maxsize check if the call to platform.architecture() doesn't work. The only issue with this is that sys.maxsize wasn't introduced until Python 2.6, so we can't use it by default yet.

-Nat

The top answer here

http://stackoverflow.com/questions/1405913/how-do-i-determine-if-my-python-s...

gets around the python 2.6 dependency.

In a nutshell:

import struct nbits = 8 * struct.calcsize("P")

James _______________________________________________ cctbxbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/cctbxbb

You are the boss! :) and if this is to deal with all the other issues - bless you mate!Cheers, o

From: [email protected] Date: Tue, 11 Jun 2013 15:11:29 -0700 To: [email protected] Subject: Re: [cctbxbb] build problem on ubuntu 12.04

On Tue, Jun 11, 2013 at 3:07 PM, Oleg Dolomanov wrote:

...

Oh, I missed an important fact - platform() MUST! return a best default, if it fails, it causes exit(1), (according to Alastair's stack trace) that cannot be caught anyway (unless you override it beforehand like sys.abort = dummy , sys.exit= dummy),

Oh, gross. What about James's suggestion? I had found that before and completely forgot, but it seems like the most universal solution so far.

-Nat _______________________________________________ cctbxbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/cctbxbb