May 2005 - Zope-Dev - Zope lists

Re: [Zope-dev] Re: Zope 2.8 on hold
by Tim Peters 08 May '05

08 May '05

[Andreas Jung] >>> At least the following are important (besides the problem with the >>> Winbuilder) [Tim Peters] >> What WinBuilders problem? I keep hearing there's "a problem" there, >> but don't know what it is -- and I had no problem using WinBuilders >> for Zope 2.8. [Andreas] > I know only that Christian had problems (according to his posting some > days ago). I can not comment and can not value these issues. But there > seems to be something. If you're talking about his posts to this list, those were all addressed some time ago, and all the problems in Zope 2.8's setup.py he was hitting have been repaired. If Christian still has problems, I haven't seen anything about them (neither on this list nor anywhere else). >> Note that these two aren't about Zope 2.8, they're about Zope 2.7.6: >> >> > http://www.zope.org/Collectors/Zope/1763 >> > http://www.zope.org/Collectors/Zope/1711 >> >> Both appear to be due to that Zope 2.7.6 on Windows shipped with a >> wrong (too old) version of pywin32, and continued to repackage pywin32 >> in the old (certifiably insane <0.7 wink>) "flat" way. >> >> Mark Hammond submitted patches to install a correct version of >> pywin32, and in a correct way. Those patches were incorporated into >> WinBuilders, so it's hard to understand how Zope 2.7.6 on Windows got >> built the way it did. Can only guess the installer was built using an >> out-of-date WinBuilders checkout, with an out-of-date pywin32, applied >> to an up-to-date Zope checkout. > Christian should know...as I said...I have no idea about how this > windows stuff works and what it does...so I depend on the competence of > you windows guys :-) It would help if Christian said something <wink>. I'm not sure the point of the above got across: the subject of this thread is "Zope 2.8 on hold", but the two issues identified above are not Zope 2.8 issues. They're Zope 2.7.6 issues. Zope 2.8 shouldn't have any problem with those. Zope-2.7.6-final-win32.exe from zope.org has those problems; they're not holding up 2.8.

3 3

ZConfig change breaks Zope 2 and Zope 3
by yuppie 08 May '05

08 May '05

Hi! On Windows, this 15 months old change in ZConfig sets the default hostname for inet_address to 'localhost': http://svn.zope.org/ZConfig/?rev=376&view=rev I have no idea why this was changed, the checkin message is not very helpful: "fix the default hostname for address datatypes for Windows" This ZConfig change is responsible for the Zope 2.8 release blocker http://www.zope.org/Collectors/Zope/1728 (the relevant info is in http://www.zope.org/Collectors/Zope/1507) and caused similar problems with Zope 3 http://www.zope.org/Collectors/Zope3-dev/383 http://mail.zope.org/pipermail/zope3-dev/2005-March/013770.html So if nobody can tell me why ZConfig was changed, I propose to 1.) revert the change and tag a new version of ZConfig 2.) update ZEO: the ZEO tests expect the old behavior, see http://mail.zope.org/pipermail/zope3-dev/2005-March/013794.html 3.) update Zope 2 trunk and Zope 3 trunk to the new versions of ZConfig and ZEO 4.) fix http://www.zope.org/Collectors/Zope/1507 and http://www.zope.org/Collectors/Zope3-dev/383 I'd volunteer to do the necessary changes in Zope 2. Any comments? Any volunteers for the other tasks? Cheers, Yuppie BTW: Why is ZConfig not a svn:external in Zope 2 and 3?

2 2

Re: SVN: Zope/trunk/lib/python/Testing/ZopeTestCase/testInterfaces.py Test interfaces with verifyClass in addition to verifyObject.
by yuppie 07 May '05

07 May '05

Stefan H. Holek wrote: > Log message for revision 30296: > Test interfaces with verifyClass in addition to verifyObject. > [...] > + self.failUnless(verifyClass(IProfiled, Profiled)) Does failUnless make any sense here? verifyClass() is either True or raises an Exception. Cheers, Yuppie

1 0

Zope 2.8, Five and Interfaces
by yuppie 06 May '05

06 May '05

Hi! I had a closer look at Zope 2.8's Five and I'm concerned about the fact that Five ships with redundant interface definitions: - redundant code is always a problem because it's hard to keep things in sync - the fact that Five is maintained in a different repository and should work with different Zope versions makes it almost impossible to change Zope interfaces in a consistent way So my questions are: 1.) Why are interfaces that are available as Zope 2 interfaces duplicated in Five/interfaces.py instead of bridged? 2.) Could we move the interfaces that are currently not available as Zope 2 interfaces to the corresponding packages in Zope 2.8, using Five/interfaces.py just as an fallback for Zope 2.7 and old Five products? If people agree that this is problem, I'd volunteer to help resolving it. Cheers, Yuppie

4 8

Using Pyrex (with BTrees)?
by Andreas Jung 02 May '05

02 May '05

Has anyone tried to use Pyrex with BTree datastructures? I tried to use Pyrex to optimize a bottleneck in TextIndexNG3 where an iteration over a list of word ids (and insertion into an IOBTree) takes very long. Unfortunately the C extension produced by Pyrex behaves differently from the original code because several unittests do fail...anyone with similar experiences? Andreas

2 1