Re: Stacking zope.component registries

On Thu, Apr 08, 2010 at 10:15:22PM -0400, Stephan Richter wrote:
> On Thursday 08 April 2010, Martin Aspeli wrote:
> > Right. However, any calls to provideAdapter() and friends would still 
> > use the global registry, unless I monkey patch 
> > zope.component.globalregistry.base, as would any ZCML directives, I guess.
> 
> That's not really monkey patching. :-) I would consider this a valid part of 
> the API. :-)

Then it's _badly-designed_ API.  Look at globalregistry.py from
zope.component 3.4.0:

  base = BaseGlobalComponents('base')

  from zope.testing.cleanup import addCleanUp
  addCleanUp(lambda: base.__init__('base'))

  globalSiteManager = base
  def getGlobalSiteManager():
      return globalSiteManager

  def provideAdapter(factory, adapts=None, provides=None, name=''):
      base.registerAdapter(factory, adapts, provides, name, event=False)

  ...

There are two names for the same thing: 'base' and 'globalSiteManager'.
Different methods use different names.  There's an assumption that you
can call __init__ on it before and after every test that uses placeless
setup.  (I hadn't noticed that last part before; I assume it'll override
base.__bases__ and destroy any attempts of stacking.)

Marius Gedminas
-- 
http://pov.lt/ -- Zope 3 consulting and development

_______________________________________________
Zope-Dev maillist  -  [email protected]
https://mail.zope.org/mailman/listinfo/zope-dev
**  No cross posts or HTML encoding!  **
(Related lists - 
 https://mail.zope.org/mailman/listinfo/zope-announce
 https://mail.zope.org/mailman/listinfo/zope )

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAku/ZywACgkQkVdEXeem149sqgCfUPkrngtUrJJi62NftNGRpF44
EssAniKVy3QTZv/AA0YAiU6YG1hTJrmS
=lsl6
-----END PGP SIGNATURE-----