- Zope-Dev - Zope lists

RE: [Zope-dev] Re: Zope 2.4.1
by sean.upton＠uniontrib.com 15 Aug '01

15 Aug '01

Where can I find the patch to Splitter.c to support [#+.] etc? There are only the changes to GlobbingLexicon.py in that thread... I second the idea that this change to ZCatalog be considered for default behavior. In order to use ZCatalog for our online classifieds search system, we had to make similar changes to Splitter.c and GlobbingLexicon's get()... Though our changes to Splitter.c were only to support smaller words so I could search for things like 'C programmer,' I wouldn't mind having the ability to search for 'C++ programmer'. Just don't try searching for 'C*' in this setup (it's mighty slow, and you shouldn't search with wildcards and a single character). Numbers should be accommodated too: a search for a 2000 Mercedes C230 ( http://classifieds.signonsandiego.com/results?searchTextWeighted=2000+Merced es+C230 ) would find matches for Mercedes, but 2000 and C230 would be stop-words in the default behavior in Splitter.c; I had to modify the source to use isalnum() instead of isalpha() so that they would not be caught as stop-words... Sean -----Original Message----- From: Chris Withers [mailto:chrisw@nipltd.com] Sent: Tuesday, August 14, 2001 8:07 AM To: Andreas Jung Cc: Tres Seaver; zope-dev(a)zope.org Subject: Re: [Zope-dev] Re: Zope 2.4.1 Andreas Jung wrote: > > > Andreas Jung wrote: > > > > > > to what fixes do you refer ? > > > > Well, at least the ones to TextIndex.py and ZCatalog.py so that > vocabularies can > > be acquired properly. > > This should work now again. Will is be in Zope 2.4.1? Not according to the placement of the Zope2.4.1-test tag in the CVS :-( Neither will the re-organisation of the Splitters to compile on Mac OS X... > > Sadly, I see Michel's patch: > > http://zope.nipltd.com/public/lists/dev-archive.nsf/ByKey/0735D7AFBA967897 > > ...never made it into the CVS :-( > > hm..I have never seen this patch before. If it is off major interest please > put it in the collector. http://classic.zope.org:8080/Collector/2514/view cheers, Chris _______________________________________________ Zope-Dev maillist - Zope-Dev(a)zope.org http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )

2 1

[Solved] Cannot load state: None has no attribute 'load'
by Dieter Maurer 15 Aug '01

15 Aug '01

Recently, I reported a strange effect where Zope suddenly and without apparent reason reports lots of "Could not load state: None object has no attribute 'load'. I think, I now understand why this happens and can work around it: We use GUF (Generic User Folder) to maintain users in an Oracle database. Although Oracle is quite fast with the kind of queries involved in authentication (with others it is extremely slow), the complete authentication process was slow: 0.2 to 0.5 seconds I wanted to improve on this and implemented a user cache. For simplicity and speed I cached the completely wrapped user object. Not a good idea, if one thinks carefully about it! Apparently, everything worked well, but suddenly, non-deterministically after many hours, Zope reports "Could not load..." in the log file and raises really erratic exceptions in Web requests.... What have I learned.... *NEVER* cache persistent objects (or objects referencing a persistent one!). They contain a hidden reference to the ZODB connection that created them (and that is used to load subobjects that have not yet been loaded). This reference is valid only in the thread (and probably also only the request) that made it. Invalid uses can show very interesting effects, like the above "Could not load...", "TypeError, 'Results' instances do not support subscription", "AttributeError, '__len__'", ..... What will I do: I will cache the bare non-persistent user object (after I scrapped all acquisition wrappers). I will rewrap it into the UserFolder context, when the cached object is looked up. Special thanks to Chris for his help with the original problem report. Dieter

2 1

HTMLDocument && infinite recursion
by Andrew T Sydelko 15 Aug '01

15 Aug '01

HTMLDocument is a product that wraps itself with the standard_html_header and standard_html_footer. Currently (version 0.3) provides no protection against recursion if either the standard_html_header or standard_html_footer are HTMLDocuments themselves. Please accept the patch at the end of this email to fix the problem. I'm cc'ing this to zope-dev to see if I'm actually fixing it in a sane way. Thanks, --andy. --- HTMLDocument.py.orig Tue Aug 14 21:59:09 2001 +++ HTMLDocument.py Tue Aug 14 22:04:36 2001 @@ -19,7 +19,7 @@ import Acquisition from DateTime.DateTime import DateTime -from AccessControl import ClassSecurityInfo +from AccessControl import ClassSecurityInfo, getSecurityManager from Globals import HTML, HTMLFile, DTMLFile, MessageDialog, InitializeClass from urllib import quote @@ -178,10 +178,15 @@ def __call__(self, client=None, REQUEST={}, RESPONSE=None, **kw): "View document" - res = self.standard_html_header(self, REQUEST) - res = res + self.body - res = res + self.standard_html_footer(self, REQUEST) - return res + security=getSecurityManager() + security.addContext(self) + try: + res = self.standard_html_header(self, REQUEST) + res = res + self.body + res = res + self.standard_html_footer(self, REQUEST) + return res + finally: + security.removeContext(self) def PrincipiaSearchSource(self): if not self.raw_text:

1 0

Re: [Zope-CMF] Re: Zope 2.4.1
by Chris Withers 14 Aug '01

14 Aug '01

Tres Seaver wrote: > > from __future__ import Zope_2_4_1 hehe... I love python > Sorry, I've been working with Brian to get it out the door, and > just jumped the gun. I should've said, "the head of the 'Zope-2_4-branch' > in CVS". Do you know if Andreas's fixes to GlobbingLexicon and the PluggableIndexes made it in? Do hope so... cheers, Chris

3 6

MySQL with Zope 2.4 & Python 2.1
by Juan Javier Carrera Obrero 14 Aug '01

14 Aug '01

Anybody is using MySQL with Zope 2.4 and Python 2.1 in Win32 plataform ? I am trying it, but after following the needed steps, that product does not run correctly. However, it runs Ok with Zope 2.1.6 and Python 1.5 version. Thanks.

2 1

manage_delObjects : problem DTML & Python
by Johan Beauce 14 Aug '01

14 Aug '01

Hello, this is my code where I have problem : <dtml-in "REQUEST.form[('idChk')]"> <dtml-call "manage_delObjects(REQUEST.idChk[<dtml-var sequence-number>])"> </dtml-in> REQUEST.form[('idChk')] return a list wich I receive from a "form". But this code is wrong because Zope don't accept a <dtml-var> in a <dtml call>. How can I insert an index for delete all objects that return my list idChk ? Do I have to use Python script for solve this problem. Thanks a lot, Johan Beaucé.

2 1

ZSQL problem when upgrading from Zope 2.3.0 to 2.3.3 (ZMySQL
by Brian E. Pearson 14 Aug '01

14 Aug '01

2.0.x to 2.0.6) Reply-To: I have searched the archives from March 2001 thru current, and could not find any information on the following problem that I am experiencing. I have actually discovered a work-a-round, but thought that I would post this message in the event others out there have experienced or are experiencing this problem. Previously when running Zope 2.3.0, and consequently a version of ZMySQLDA prior to 2.0.4, I was able to display the values of int(11) fields from MySQL without any problem. However, upon upgrading to Zope 2.3.3 and ZMySQLDA 2.0.6, this broke. All of my DTML-Methods and Documents were now displaying the long integer indicator "1L", "2L", etc. Additionally, other ZSQL methods that I had created which took integer arguments were now breaking as they were trying to query the database with values of "1L", "2L", etc. rather than 1, 2, etc. The solution: According to the CHANGES document for ZMYSQLDA: As of ZMYSQLDA 2.0.4, all integers are returned as long integers in Python. Unfortunately, Python 1.5.2 appends an "L" to these values when rendering them. To resolve the formatting/printing/display issue, either: a.) Upgrade to Python 2.0 or greater b.) Use a format string on the integer columns <dtml-var varname fmt="%d"> To resolve the ZSQL methods from blowing up either: a.) Upgrade to Python 2.0 or greater b.) Use <dtml-sqlvar varname type=int> instead of <dtml-var varname> in your ZSQL methods Hope this helps someone else. -Brian

1 0

Re: [Zope-dev] Status of ZopeStartupProvisions?
by Anthony Baxter 14 Aug '01

14 Aug '01

>>> "R. David Murray" wrote > I wasn't aware of that Wiki, or of zctl.py until after I wrote something > of my own along these lines: a 'zopectl' script that has a slightly > different design objective than zctl.py. zctl.py seems to be oriented > primarily toward managing ZEO instances. zopectl is oriented toward > managing multiple INSTANCE_HOME setups running on the same box > (eg: production and test instances, migration instances, or indeed > multiple zope-instance-per-virthost virthosts). > You'll find it at http://www.zope.org/Members/rdmurray/zopectl. I'll have to have a look at it - I've built my own tool to manage multiple ZEO instances across multiple hosts, as well as allowing upgrades from development -> internal test -> external test -> production along with starting up temporary servers and frobbing loadbalancers to make sure upgrades are transparent to the end users. This is probably a tad overkill for most people, though. Anthony

2 1

Status of ZopeStartupProvisions?
by Richard Jones 14 Aug '01

14 Aug '01

We're currently trying to neaten up Zope startup (zope startup for dummies is currently a real pain). I ran across the ZopeStartupProvisions wiki, and it's pretty close to what we're aiming for - having a much nicer startup/control script and a configuration file. It's been dormant since April, though, and I'm wondering if there's still anyone backing it? Richard

2 1

Re: Zope-Dev digest, Vol 1 #1267 - 9 msgs
by Ken Manheimer 13 Aug '01

13 Aug '01

I'm unable to reproduce the CVS problems you both report! Specifically, i tried a checkout and build this morning: cvs -d :pserver:anonymous@cvs.zope.org:/cvs-repository co Zope and was able to build and run without any failures. Further, 'grep' for regex in lib/python/DateTime: grep regex * */* turned up no results. Just to eliminate possible missteps, could you check some things? - Check that the contents of CVS/Repository in the homedir of the checkout contains:: Zope It would be worth confirming that the CVS/Root file contains:: :pserver:anonymous@cvs.zope.org:/cvs-repository and that there is no file named 'Tag' in that directory. - If any of those don't match what they should be, redo the checkout, and check whether you're still getting the bogus stuff. - Confirm that cvs.zope.org resolves to 63.102.49.30. ('host cvs.zope.org') Let me know what you find... Ken klm(a)zope.com On Sun, 12 Aug, Tino Wildenhain <tino(a)wildenhain.de> wrote: > I saw many removements of code. Very strange. Also the HEAD > tag was screwed apparently. > > Hope Brian can tell. > > --On Sonntag, 12. August 2001 14:15 +0200 Dieter Maurer <dieter(a)handshake.de> wrote: > > > A few days ago, I made a fresh checkout of module "Zope" > > from ":pserver:anonymous@cvs.zope.org:/cvs-repository". > > > > I had hoped, I would get quite new Zope sources, something like 2.4+. > > But apparently, the available sources there are old > > and inconsistent. > > > > Up to now, I met several problems: > > > > 1. "cPickle.c" wants to include non-existent "mymath.h" > > > > 2. "z2.py" tries to import "ZLogger", but this is package > > "Zope" (ImportError) > > > > 3. "DateTime" still uses "regex" instead of "re" > > > > 4. "FileStorage" calls "append" with 2 rather than 1 argument > > > > 5. The documentation tells: you need "Python 1.5.2". > > > > What goes on?

3 4