[Zope3-dev] Re: 'thread_local' AttributeError in zope.security.checker

[Tres Seaver]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jim Fulton wrote:
> On Aug 14, 2007, at 7:32 PM, Paul Carduner wrote:
> 
>> Hello zope3-dev,
>>
>> I came across an error I have not seen before.  Just for reference,
>> here is the last snippet from the traceback:
>>
>> File "/home/pcardune/Work/ZContact/zcontact-lp/eggs/tmpgwuq6O/ 
>> zope.security-3.4.0b4-py2.4-linux-i686.egg/zope/security/checker.py",
>> line 420, in ?
>> AttributeError: 'module' object has no attribute 'thread_local'
> 
> I wish you had included the whole traceback.
> 
> 
>> The interesting part to me is that it was so easy to fix.  Here is a
>> snippet of the relevant area of failing code from
>> zope/security/checker.py
>>
>> # Get optimized versions
>> try:
>>     import zope.security._zope_security_checker
>> except ImportError:
>>     pass
>> else:
>>     from zope.security._zope_security_checker import _checkers,  
>> selectChecker
>>     from zope.security._zope_security_checker import NoProxy, Checker
>>     from zope.security._zope_security_checker import _defaultChecker
>>     from zope.security._zope_security_checker import  
>> _available_by_default
>>     zope.interface.classImplements(Checker, INameBasedChecker)
>>
>> The very first import statement is throwing the error, but alas, it is
>> not an ImportError.  If I patch this code by catching the
>> AttributeError in the same way the ImportError is caught, then
>> everything works fine.
> 
> Except you don't get the C optimizations. IOW, you would just be  
> covering up a real error.
> 
> 
>> I tried looking into the c code, but after
>> years of python development, the c code just frightens me (a bit).
>> That said, I would really appreciate any pointers people might be able
>> to give me on what thread_local is for and why I (don't) need it.
> 
> I rearranged the imports in zope.security yesterday.  I apparently  
> broke your usage, but I can't tell what your usage is.  I'll need to  
> see the whole traceback.  In particular, I'll need to see the import  
> order.

We badly need to get a handle on how we expect people to fetch eggs for
packages:  having such a refactoring affect an indirect client (one not
expecting to track day-to-day development) is a disaster.

Dumping revision-stamped (unreleased) eggs into a directory used by
unsuspecting users is a recipe for bringing *all* development to a
screeching halt.


Tres.
- --
===================================================================
Tres Seaver          +1 540-429-0999          tseaver at palladion.com
Palladion Software   "Excellence by Design"    http://palladion.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGww8E+gerLs4ltQ4RAizUAJ4/g04M14f9pI2QA/NAub2CwrVtDwCfQ2cD
dQNNFuvkirf3sbAWOuC7/VE=
=YjyB
-----END PGP SIGNATURE-----