- Zope-Dev - Zope lists

Making Role.py's __ac_local_roles__ play nicely with computed attributes
by Christian Theune 25 Nov '04

25 Nov '04

Hi, we currently have a project where we calculate local roles dynamically. This involves using a computed attribute as an __ac_local_roles__ replacement. Unfortunately this doesn't mix too well with assigning them manually, although our computed attribute takes care. Why is this and what can we do against that? __ac_local_roles__ is a dict. It's not a PersistentDictionary, but an ordinary Python dict. Therefore, to obey the rules of persistence, the Role.py local role machinery reassigns __ac_local_roles__ after every change. I'd like to change this, so that Role.py sets _p_changed instead, which is what it tries to signal. Additionally you need to explicitly assign __ac_local_roles__ instead of just getting dict=self.__ac_local_roles__ or {} so this becomes dict=self.__ac_local_roles__ if dict is None: self.__ac_local_roles = dict = {} This change works without disrupting the unit tests and I would love to get it on Zope 2.8. Any objections? I take silence as a 'no objections'. Cheers, Christian -- gocept gmbh & co. kg - schalaunische str. 6 - 06366 koethen - germany www.gocept.com - ct(a)gocept.com - phone +49 3496 30 99 112 - fax +49 3496 30 99 118 - zope and plone consulting and development

2 1

Zope Community Contentment (Part II)
by Stephan Richter 23 Nov '04

23 Nov '04

Hi everyone, a while back Katrin Wodzicki (Katrin.Wodzicki_at_uni-jena.de) conducted a poll about the contentment of the Zope community. She has now prepared a second questionnaire that is based on the answers from the first one. This questionnaire is pretty short and I filled it out in about 10 minutes (since I wrote a lot -- most people will need only 5 minutes). The questionnaire is available bi-lingual in German and English at http://evaluation.gocept.com until December 5. Please take a moment to fill out the forms. It will help not only Katrin to finish her Diploma (German College degree), but also the Zope Community to better understand our community's strongholds and shortcommings. The results will be presented at the Sprint DZUG meeting and will be available afterwards online. So please, hit the server with some traffic! Regards, Stephan -- Stephan Richter CBU Physics & Chemistry (B.S.) / Tufts Physics (Ph.D. student) Web2k - Web Software Design, Development and Training

1 0

[Caching] Inefficient HTML cache?
by Dieter Maurer 23 Nov '04

23 Nov '04

We use Apache for caching and observe the following behaviour: Apache requests the object from Zope and caches it until the expiration date is reached. From then on, Apache forwards "If-Modified-Since" requests to Zope, Zope responds with an "304" response but without a new expiration date! As Apache does not get new expiration info, it continues to forward "If-Modified-Since" requests for the object. This significantly reduces cache efficiency. In my view, Zope should include new expiration info in its "304" responses. The CachingManager API may need to be extended, to allow the HTTP CacheManager to add information to "304" responses. -- Dieter

2 2

Re: [Zope-Checkins] CVS: Zope - test.py:1.2.2.16
by Stefan H. Holek 22 Nov '04

22 Nov '04

Sorry Florent, I missed your warning mail. I have two issues with the patch: a) Trailing slash no longer works (--libdir Products/) b) Some combinations of --libdir and --dir and symlinks no longer work on Mac OS X (HFS) Unfortunately os.getcwd() returns different paths in the presence of symlinks on Mac and Linux (already realpath'd vs not). I don't have a good solution for this either. You have a Mac haven't you? Does it work for you across all platforms? Thanks, Stefan On 19. Nov 2004, at 14:49, Florent Guillaume wrote: > Update of /cvs-repository/Zope > In directory cvs.zope.org:/tmp/cvs-serv20719 > > Modified Files: > Tag: Zope-2_7-branch > test.py > Log Message: > Fixed to work in the presence of symliked products or Products > directory. > > > === Zope/test.py 1.2.2.15 => 1.2.2.16 === > --- Zope/test.py:1.2.2.15 Sat Oct 30 04:29:22 2004 > +++ Zope/test.py Fri Nov 19 08:48:59 2004 > @@ -380,11 +380,12 @@ > self.cwd = os.path.realpath(os.getcwd()) > # Hack again for external products. > if libdir: > - self.libdir = os.path.realpath(os.path.join(self.cwd, > libdir)) > + self.libdir = os.path.join(self.cwd, libdir) > else: > - self.libdir = os.path.realpath(os.path.join(self.cwd, > self.libdir)) > - if self.libdir not in sys.path: > - sys.path.insert(0, self.libdir) > + self.libdir = os.path.join(self.cwd, self.libdir) > + real_libdir = os.path.realpath(self.libdir) > + if real_libdir not in sys.path: > + sys.path.insert(0, real_libdir) > # Determine where to look for tests > if test_dir: > self.testdir = os.path.abspath(os.path.join(self.cwd, > test_dir)) > > _______________________________________________ > Zope-Checkins maillist - Zope-Checkins(a)zope.org > http://mail.zope.org/mailman/listinfo/zope-checkins > -- The time has come to start talking about whether the emperor is as well dressed as we are supposed to think he is. /Pete McBreen/

2 2

Zope2 OFS/CopySupport's _get_id()
by Alexandre Boeglin 22 Nov '04

22 Nov '04

Hello, While I was reading the _get_id function, I felt it was really strange : #### def _get_id(self, id): # Allow containers to override the generation of # object copy id by attempting to call its _get_id # method, if it exists. n=0 if (len(id) > 8) and (id[8:]=='copy_of_'): n=1 orig_id=id while 1: if self._getOb(id, None) is None: return id id='copy%s_of_%s' % (n and n+1 or '', orig_id) n=n+1 #### currently, when copying and pasting an object, 'copy[0-9]*_of_' will be prepended to the id (if the id is already in use in the current folder). Thus, copying 'x' will create 'copy_of_x', 'copy2_of_x', etc. and copying 'copy_of_x' will create 'copy_of_copy_of_x', 'copy2_of_copy_of_x', etc. What I don't get, is what the two lines between 'n=0' and 'orig_id=id' are supposed to do. What they do is this : if an object id is 16 chars long, and it ends with 'copy_of_' (like 'XXXXXXXXcopy_of_'), its first copy will be called 'copy2_of_XXXXXXXXcopy_of_' instead of 'copy_of_XXXXXXXXcopy_of_'... What I think the author intended to do, is that, when you copy an object called 'copy_of_x', the copy will be named 'copy2_of_x' instead of 'copy_of_copy_of_x'. Thus, here's a function that does the job : #### import re copy_re=re.compile('^copy[0-9]*_of_') def _get_id(self, id): # Allow containers to override the generation of # object copy id by attempting to call its _get_id # method, if it exists. copy_match=self.copy_re.match(id) if (copy_match) and (copy_match.end() < len(id)): n=1 orig_id=self.copy_re.sub('', id) else: n=0 orig_id=id while 1: if self._getOb(id, None) is None: return id id='copy%s_of_%s' % (n and n+1 or '', orig_id) n=n+1 #### Then, I don't know what is the preferred naming for new objects, but I fear that 'copy_of_copy_of_copy_of_copy_of_copy_of_copy_of_object' will quickly be too long, if objects are copied by an automated process that doesn't show the Ids to the user or don't allow their modification. What are your feelings about this ? Regards, Alexandre Boeglin

2 1

re: [Zope-dev] Zope on 64-bit Linux
by John Snowdon 22 Nov '04

22 Nov '04

Hi Larry, We are using Zope/ZEO 2.7.x on a small cluster of AMD Opteron machines to host an online learning environment for our medical students (around 1100 in all), providing on-the-fly timetables, study and lecture notes, module booking, downloadable resources, discussion forums, etc.. Machines are Sun V20z dual 248 Opterons, each with 4Gb of RAM, dual gigabit ethernet and 10krpm scsi disks. This is all hosted on a private gigabit network, with a local LDAP authentication server. Access to the sites is via a load-balancing gateway running 'Pound'. The Opteron nodes (3 of the above spec, at the moment) are running SuSE 9.1/Pro/X86_64, with the SuSE provided python64 binaries and libs. All the extensions and py modules we use are compiled against /lib64. The Opteron nodes host the sites, using Zope 2.7.2, connecting to a bigger, but slower Sun fileserver, which serves the data fs via ZEO. Performance is blistering; but we could easily improve it with squid. We've benchmarked the Opterons against dual Xeon systems (2.8, 3.0) and they are *way* faster. If you want any more details, just let me know... >Anyone out there successfully deploying and running Zope/CMF on 64-bit >Linux? I'm looking at deploying AMD Opteron server for hosting large >Zope site and wanted to take advantage of the speed and addressing >capabilities if I could. Specifically I'd like to use RedHat Fedora >Core 3 to run ZODB server and ZEO Clients. Any information would be >greatly appreciated. > >Thanks in advance, >Larry Bates John Snowdon - IT Support Specialist -==========================================- School of Medical Education Development Faculty of Medical Sciences Computing University of Newcastle Phone : 0191 246 4549 Email : j.p.snowdon(a)ncl.ac.uk

1 0

Zope on 64-bit Linux
by Larry Bates 19 Nov '04

19 Nov '04

Anyone out there successfully deploying and running Zope/CMF on 64-bit Linux? I'm looking at deploying AMD Opteron server for hosting large Zope site and wanted to take advantage of the speed and addressing capabilities if I could. Specifically I'd like to use RedHat Fedora Core 3 to run ZODB server and ZEO Clients. Any information would be greatly appreciated. Thanks in advance, Larry Bates

2 1

exception hiding in _delObject
by Florent Guillaume 19 Nov '04

19 Nov '04

In _delObject and manage_beforeDelete there's a try/except that catches nearly everything (except BeforeDeleteException and the infamous ConflictError). It then proceeds to log a message, but continues: try: object.manage_beforeDelete(object, self) except BeforeDeleteException, ob: raise except ConflictError: # Added for CPS raise except: LOG('Zope',ERROR,'manage_beforeDelete() threw', error=sys.exc_info()) pass The rationale is, I guess, that when you delete an object you really want it gone. This is IMO very harmful because it hides any bug in the catalog or the indexes (especially during unit tests where LOG is ignored). I've been (again) bitten by it. I'd like to condition the pass to the fact that the current user is Manager. Otherwise I'd like it to fail (and reraise). So a Manager will still be able to delete objects when there's a bug, but not others. Comments ? Florent -- Florent Guillaume, Nuxeo (Paris, France) CTO, Director of R&D +33 1 40 33 71 59 http://nuxeo.com fg(a)nuxeo.com

3 5

Announce: November Zope Bug Day!
by Chris Withers 19 Nov '04

19 Nov '04

Hi All, Well, October came and went with no bug day. Apologies to all :-S However, the November bug day will shortly be upon us! What? Zope Bug Day working from http://collector.zope.org/Zope When? Friday 26th November - from when you wake up until when the day ends! Where? #zope-dev on irc.zope.org Come along and help if you've got a chance! Read this if you want to know how: http://dev.zope.org/CVS/BugDays cheers, Chris -- Simplistix - Content Management, Zope & Python Consulting - http://www.simplistix.co.uk

2 1

Re: [Zope-Checkins] CVS: Packages/ZPublisher - Converters.py:1.23.2.3 HTTPRequest.py:1.90.2.7 HTTPResponse.py:1.75.2.7
by Chris Withers 19 Nov '04

19 Nov '04

Andreas Jung wrote: > Update of /cvs-repository/Packages/ZPublisher > In directory cvs.zope.org:/tmp/cvs-serv30970/lib/python/ZPublisher > > Modified Files: > Tag: Zope-2_7-branch > Converters.py HTTPRequest.py HTTPResponse.py > Log Message: > 'latin1' -> 'iso-8859-15' > > > === Packages/ZPublisher/Converters.py 1.23.2.2 => 1.23.2.3 === > --- Packages/ZPublisher/Converters.py:1.23.2.2 Mon May 17 13:31:31 2004 > +++ Packages/ZPublisher/Converters.py Mon Nov 8 01:49:18 2004 > @@ -20,7 +20,7 @@ > def field2string(v): > if hasattr(v,'read'): return v.read() > elif isinstance(v,UnicodeType) : > - return v.encode('latin1') > + return v.encode('iso-8859-15') > else: > return str(v) Should this really be hard coded at all? http://collector.zope.org/Zope/1490 Chris -- Simplistix - Content Management, Zope & Python Consulting - http://www.simplistix.co.uk

1 0