June 1999 - Zope-Dev

[ANN] Zope in the low countries
by Rik Hoekstra 09 Jun '99

09 Jun '99

I proudly announce the birth of the usergroup for Dutch speaking Zope users. As they will probably mostly be from Belgium and the Netherlands (though the python newsgroup shows you never know :-) I called it Zope in the low countries (in dutch: Zope in de lage landen) The most important aim of the usergroup are: - to provide mutual assistence for Zope users on a regional level, - to be able to provide institutional decision makers with a guarantee that Zope is not some obscure and unreliable open source product but has more users and (hopefully) a great many succesfull implementations nearby. There is a discussion list coming (ZOPE-NL), but not yet operational. Details will be on the web site. To keep things focused, ZOPE-NL will be no local replacement for the regular zope lists, but a discussion forum for regional matters only. For the time being, if you're interested, please visit the (rudimentary) site at http://sw.fsw.leidenuniv.nl/cgi-bin/WebSite.exe/zopegroep Warning: it is completely in dutch. Rik Hoekstra

1 0

XML-RPC and Zope
by Eric Kidd 08 Jun '99

08 Jun '99

Hello! I'm back, after a brief leave of absence. If anybody out there is working on XML-RPC support for Zope, I'd like to speak with you. I took a stab at this a while back, but ran out of time. Cheers, Eric

1 0

DA changes needed for Zpe 2
by Jim Fulton 08 Jun '99

08 Jun '99

Thread safety of DAs will be a significant issue in Zope 2, since Zope 2 wants to be multi-threaded, although it can still be run in a multi-threaded mode. Here's the scoop: If an underlying database API is thread safe and allows multiple connections in multiple threads, then nothing needs to be done. Database connections will be automatically pooled as part of the Zope database connection pooling mechanism. RDBMS database connections will be established as needed and kept only as long as needed to satisfy load. This will be very cool. Unfortunately, I fear that many of the database APIs (or possibly the Python database interfaces to them) are not thread safe. In the short term, I'll come up with a mechanism for assuring that only one thread is using a database at one time. This will allow some (most?) databases to limp along. I fear that some DAs will never work with a multi-threaded zope due to limitations in the underlying APIs. There are probably a number of DAs that can and should be updated to work correctly in a multi-threaded environment, in which multiple threads may have connections to the same database. People working on DAs should start thinking about this. Jim -- Jim Fulton mailto:jim@digicool.com Technical Director (540) 371-6909 Python Powered! Digital Creations http://www.digicool.com http://www.python.org Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.

1 0

Re: [Zope-dev] Q/ regarding 'Broken Product'
by Jim Fulton 08 Jun '99

08 Jun '99

Thomas Riedl wrote: > > Ah, I like turn around THAT short. > You guys at DigiCool are simply great! I try to pay close attention to anything having to do with the database. :) > .... > >> ======================= > >> Whenever there's a "severe" error (like a syntax error in the Product > >> python code), I see a 'broken Product'. > >> > >> Sometimes I'm able to fix the error and to then reuse my created Product > >> instances and all its subsidiary objects. > >> > >> Sometimes the Product instance will have ceased to exist, which is quite a > >> P.I.T.A. > > > >I've never seen this. It would be interesting to use the bbb.py script in > >the utilities directory (of Zope 2.0a1 or in the public CVS) to create an XML > >rendition of your database. This might give you a better idea where your data > >has gone. :) > > I tried, but it breaks on importing 'ppml' when I try to report xml. See > below... Is there a ppml.py in your utilities directory? If not, I can send it to you, however, I see what your problem is wrt disappearing instances. See below. > >Uh, what exactly do you mean by ceased to exist? What is your evidence > >for this? Do they no longer show up in otherwise functioning folders? > > My evidence looks like this: > I have my Product, XXX. > I instatiate it. > It works. > I break it (by e.g. syntax errors). > I fix them. > I change into the /Control Panel/Products > folder; there's then my still broken Product. > clicking on it sometimes reveals an instance (created earlier via selecting > 'XXX' from the selection list, clicking 'Add', a.s.o.), and sometimes > that (or those) instance(s) aren't visible any longer. > The XXX Folder reads > 'There are currently no items in Broken product XXX' > and offers the 'add something' selection list. > > I believe to have found out this pattern: > Especially tricky seem to be changes with / deletions of method names, but > sometimes a carefully left out (;-) paren will do, as long as the error > occurs in Zope's startup phase. Ah. The fact that Zope lets you add items to a Python product (as apposed to a through-the-web product) is a bug. I'll add this to the bugs list. When a Python-based product changes state from broken to OK or OK to broken, the Zope product object gets recreated. Add your instances to a folder in the main tree or in a through-the-web product. > >> Nevertheless, the Product keeps on showing up as 'Broken', including its > >> out-dated Traceback, until I "delete" it and restart Zope. > > > >What version of Zope are you using? I thought 1.10.2 got this right. > > It is 1.10.2, running on a Linux RedHat-5.2 box, python-1.5.1, using > Zope's own HTTP server (like "python serve.py"). Hm. Waaaa. I know this is fixed in the current code. > >> Q#1: How can I 'mend' a broken Product w/o deleting it? > > > >You *should* be able to just mend it. If it doesn't show up > >as mended, then there is a Zope bug. > > Well, then I regard this an undocumented feature... I think it > doesn't stem from my browser caching data, since I changed its amount > of cache down to 2k (you know, Netscape and its 'caching' of erroneous > pages by the last error-free content ...). I'm guessing that it is a bug in 1.10.2. This is fixed in 2.0. In the mean time, you will have to work around the problem by moving (deleting) the product. Note that another work-around may be to modify (create) a version.txt in the product folder. Also note that, even through the product shows up as broken, you may still have a functioning product. Do your product items show up in the add list? > >> Q#3: When I change attributes of e.g. an anonymous keyword object: will > >> this inevitably result in the often_heard_of database > >> inflation 'til next pack day? > > > >I'm not sure what you are refering to here. When you change any > >(persistent) object, a new version of the object is written to the database. > >The old version is retained until it is packed. > > That's what I intended to ask for. > > So, what shall I do upon the next mysterious disappearance on instances? See above. Don't poke instances into the product folder (of a Python written product.) Jim -- Jim Fulton mailto:jim@digicool.com Technical Director (540) 371-6909 Python Powered! Digital Creations http://www.digicool.com http://www.python.org Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.

2 1

Question to acquisition
by Thomas Riedl 08 Jun '99

08 Jun '99

Hi again; after some source code studies I found out that I can return filled-out DTML from my python code by calling the returned Doc.Temp.: ... template=DocumentTemplate.HTMLFile(""" """, globals()) ... do something, determine value for myvarname return template(globals(), myvarname=determined_result, some_name='hello') But as soon as I want to use inherited attributes (like e.g. standard_html_header), I get a keyword error from out of DT_String.py. I saw that I can specify as well a mapping as a client when calling the template, but I do not know what to specify for each to get what I want: / standard_html_header <----- This Me Wants !! Products / MyProd / MyInstance / some_python_code <--- Therein I want to call: template(what? , what?, globals(), myvarname=3) Any clues? -- MfG: Thomas Riedl =========================================================================== triedl(a)linuxland.de, (+49 89) 99 34 12 - 11, http://www.linuxland.de Stefan-George-Ring 24, 81929 München

2 1

Q/ regarding 'Broken Product'
by Thomas Riedl 08 Jun '99

08 Jun '99

Hi there; stumbling over a $§#@ issue, a few questions arose regarding Products: I'm happily zopeing away my time while trying to implement a simple Product; this is group of a keyword-indexed list of incidents; one could easily think of it as some kind of support system. I started to implement this as 3 python classes, namely the group, the keyword and the incident class; the group objects are added via selection list and Add button, so they have Zope-ID's. They contain lists of keyword and incident object instances, with the keyword and incident objects having their own reference lists. All my objects are created persistent. Problems: ======================= Whenever there's a "severe" error (like a syntax error in the Product python code), I see a 'broken Product'. Sometimes I'm able to fix the error and to then reuse my created Product instances and all its subsidiary objects. Sometimes the Product instance will have ceased to exist, which is quite a P.I.T.A. Nevertheless, the Product keeps on showing up as 'Broken', including its out-dated Traceback, until I "delete" it and restart Zope. Q#1: How can I 'mend' a broken Product w/o deleting it? Q#2: How can I avoid the grim reaping of its instances upon tiny errors like a missed out ')' ? Q#3: When I change attributes of e.g. an anonymous keyword object: will this inevitably result in the often_heard_of database inflation 'til next pack day? Flame bait: Q#4: Am I doing The Right Thing (creating 'anonymous' objects), or would a Zopeing One Perfectly Enlightened choose to e.g. create tons of directly accessible objects like 'ID=keyw_adaptec_scsi_controller' or 'ID=incident_1999_05_03_17_42_05_923' ? Currently I'm quite happy using methods mapping object identifiers to objects (like ) and then accessing their attributes 1K TIA tom (in heavy reconsideration mode right before the final coding) =========== MfG: Thomas Riedl =========================================================================== triedl(a)linuxland.de, (+49 89) 99 34 12 - 11, http://www.linuxland.de Stefan-George-Ring 24, 81929 München

2 1

handling exception in dict lookup during URL untangling; and sessions
by Michael Olivier 06 Jun '99

06 Jun '99

Hi folks, I'm subclassing the BoboDictionary class from a while ago to do sessions (not versions, but tracking user sessions at the web site). I cannot assume cookies are available for my users, so I'm using a session code as part of the URL, e.g., localhost/sess/RQZ/someObject, where RQZ is the session ID. def __getitem__(self,key): # override this one function of BoboDictionary try: session = self.buffer[key] session.accessed() return session except: raise DMSessionMgrSessionExpired The problem I'm having is that when the session times out, my code raises a DMSessionMgrSessionExpired exception which I haven't figured out how to trap somewhere because it happens during Zope's untangling of the URL. I can't return a web page from this function (__getitem__()). I'd like it to show an explanatory (error) page to the user, with a link to login again. I could create a "dummy" session ID, which __getitem__() would return, and then all functional entry points would have to recognize that dummy ID and go to the error page. That's pretty painful. Is there a better approach I could take? And re. sessions in general -- is sessions support coming in Zope? I have code I could contribute along these lines if it would help. thanks! Michael

1 0

Persistence and __getattr__
by kent＠tiamat.goathill.org 05 Jun '99

05 Jun '99

It appears to me that if you have a class object that you want to be Persistence, all the object attributes have to be in object.__dict__ or Persistence.__getattr__ has no way of retrieving them. This at first seems pretty obvious, but what about instances where your object attributes change outside of zope? I can't determine any way to trigger an 'attribute refresh' on a __getattr__ failure if the attribute does not exist in the object.__dict__. Is this correct? Is it that 'volatile' attributes cannot be Persistent? I see that when I try to override Persistence.__getattr__ in order to add the capability to create or change attributes while still seemingly duplicating the the method instructions, I encounter an infinite loop. :^) Yes, I read the docs, but I'm trying to figure out if there is perhaps a way around this puzzle. Thanks Kent

4 3

Attribute Error in _call_ with childless_decoration
by Jason Spisak 01 Jun '99

01 Jun '99

Zopeinators: It worked fine, then all of a sudden, I got a traceback on a document that uses this code:  <a href=""> <img src="" border="0">   </a>  It worked Friday, I didn't do any work over the weekend, and now it gives me this traceback:  Traceback (innermost last): File /home/jason/Zope/lib/python/ZPublisher/Publish.py, line 255, in publish_module File /home/jason/Zope/lib/python/ZPublisher/Publish.py, line 157, in publish File /home/jason/Zope/lib/python/ZPublisher/mapply.py, line 154, in mapply (Object: index_html) File /home/jason/Zope/lib/python/ZPublisher/Publish.py, line 98, in call_object (Object: index_html) File /home/jason/Zope/lib/python/OFS/DTMLMethod.py, line 157, in __call__ (Object: index_html) File /home/jason/Zope/lib/python/OFS/DTMLMethod.py, line 153, in __call__ (Object: index_html) File /home/jason/Zope/lib/python/DocumentTemplate/DT_String.py, line 513, in __call__ (Object: index_html) File /home/jason/Zope/lib/python/TreeDisplay/TreeTag.py, line 154, in render (Object: a tree tag) File /home/jason/Zope/lib/python/TreeDisplay/TreeTag.py, line 244, in tpRender (Object: 1000059) File /home/jason/Zope/lib/python/TreeDisplay/TreeTag.py, line 481, in tpRenderTABLE (Object: 1000059) (Info: ([113750, []], {'childless_decoration': '', 'id': 'tpId', 'branches_expr': <method Eval.eval of Eval instance at 9c2a9a8>, 'url': 'tpURL'}, ([113750, []],), ([113750, []],))) File /home/jason/Zope/lib/python/TreeDisplay/TreeTag.py, line 400, in tpRenderTABLE (Object: 1000061) File /home/jason/Zope/lib/python/DocumentTemplate/DT_With.py, line 132, in render (Object: Candidates) AttributeError: (see above) Any ideas what a childless_decoration is? The 'Submission' object is a class that has a property named 'candidate' which containes the id of an object in the 'Candidates' folder. I'm not sure what's going on. Any insight is greatly appreciated. Thanks in advance, Jason Spisak webmaster(a)mtear.com

2 1

Re: [zope-dev] zope/apache - external product writing to a file system ?
by PP Jose C. Lacal 01 Jun '99

01 Jun '99

When will you release the product? Also, what about the opposite effect: - An external method that **reads** existing HTML files from the file system? The possible application? Let�s say I host my website with a hosting service that does not support Zope directly. Yet, if I am able to install Zope on my personal disk space, I could get Zope running (on port :81, say) and then Zope would read and write the HTML files to the filesystem. The point: using Zope to *maintain* my site, through a web interface that is a lot friendlier than FTPing HTMl files up to the web hosting service. Just an idea. Regards. --- "Dr. Ross Lazarus" <rossl(a)med.usyd.edu.au> wrote: > > > > > We're running zope 1.10.2 under apache. > > Briefly, my question is - What's the easiest way to > convince apache/zope > to allow an external method to actually write some > files somewhere - do > I only need to ask my friendly (!) sysadmin to > adjust some apache > permissions or is there more to this ? > > Details - In order to get a static html > representation of a zope folder > and subfolder objects, I thought I'd write myself an > external method > which walks the folder structure and tries to write > the dtml stuff to > the file system as a static mirror of the dynamic > zope site. > > Problem. Everything seems to work fine except that > all of the file > writes fail - IOError 13 - permission denied. > > I thought I'd be sneaky and tried using the zope/var > directory assuming > that apache was stopping zope from writing to any > arbitrary directory. > No joy - zope too smart to allow me to do that. > > > I'll release it as a contributed product if I can > get this sorted out. > (Yes, wget already does this but we've had some > interesting problems...) > > -- > > Dr Ross Lazarus > Associate Professor and Sub-Dean for Information > Technology > Faculty of Medicine, A27, R126a, University of > Sydney, > Camperdown, NSW 2006, Australia > Tel: (+61 2) 93514429 Mobile: +61414872482 > Fax: (+61 2) 93516646 Email: rossl(a)med.usyd.edu.au > http://www.health.usyd.edu.au/people/rossl.htm > > _______________________________________________ > Zope-Dev maillist - Zope-Dev(a)zope.org > http://www.zope.org/mailman/listinfo/zope-dev > > (For non-developer, user-level issues, use the > companion list, > zope(a)zope.org, > http://www.zope.org/mailman/listinfo/zope ) > > > > > === = = = = = \ / = = = = = / \ = = = = = [PP] Jose C. Lacal jose.lacal(a)volks-pc.org Visit my website at http://www.volks-pc.org _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com

4 3