[AG-TECH] Building failure - strange behaviour with pyGlobus

Joseph joseph at cs.usyd.edu.au
Mon Sep 6 18:29:01 CDT 2004


Hi

This seems to be the solution, however I can't test it today and I will  
leave soon. But I will upgrade the documentation and tell Steve who will  
continue this work.

Thanks a lot for your answers.

Cheers,
	Joseph

On Mon, 06 Sep 2004 10:58:48 -0500, Thomas D. Uram <turam at mcs.anl.gov>  
wrote:

> Chris:
>
> You're right, it likely has to do with the change in flavor.  Lately  
> I've come to understand that we should only be using debug-built globus,  
> hence the change to gcc32dbgpthr.
>
> Joseph:  If you build globus with flavor gcc32dbgpthr, are these  
> problems resolved for you?
>
>
> Tom
>
>
> Chris Willing wrote:
>> On Fri, 2004-09-03 at 15:34, Joseph wrote:
>>
>>> Hi
>>>
>>> In fact, since the Beta2 release, my build isn't working anymore. The   
>>> pyGlobus module isn't build. In my build output, I find a lot of :
>>> "Building pyGlobus
>>> Checking for which modules to build
>>> Checking the util module
>>> Missing dependency to build the util package.
>>> This package is required to build all of pyGlobus.
>>> The following is the error message from globus-makefile-header.
>>> ('No packages were found that matched your query!\n',)
>>> Now exiting from the build."
>>> And so on, then
>>> "ImportError: No module named pyGlobus"
>>>
>>> Nothing had changed between the Beta1 and the Beta2 on my pc, so I  
>>> filed a  bug report but apparently it wasn't one, the trouble should  
>>> be with my  installation of environment variables.
>>>
>>   I believe this is due to a change in packaging/BuildPythonModules.py,  
>> in
>> which the value of the flavor argument when building pyGlobus for linux
>> has recently changed from 'gcc32pthr' to 'gcc32dbgpthr' (at line 71).
>>  Reverting to
>>   flavor = 'gcc32pthr'
>> allows a build to complete and allows the post installation script
>> AccessGrid-Postinstall.py to complete without error. It also allows
>> certmgr.py to run again, so I suspect it will allow VenueClient to run
>> again too (all of these previously had the same run time error about
>> lack of pyGlobus), but I'm currently not anywhere I can test that till
>> tomorrow morning.
>>  Still, I guess there was some reason to change flavor's value from
>> 'gcc32pthr' to 'gcc32dbgpthr', which a reversion won't help.
>>  chris
>>



-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/m2/




More information about the ag-tech mailing list