- Zope-Dev - Zope lists

anyone got any code for dealing with batches and LazyFilter?
by Subjectivize 15 Nov '01

15 Nov '01

I'm using a ZCatalog/LazyFilter inside a dtml-in method. Unfortunatly when in comes to batching etc the result is that the list is not longer lazy. Everything gets evaluated. Anyone come accross a similar problem and found a "nice" way of dealing with it? What is needed is something that uses the start index of the underlying list rather the filtered one but then I'm not sure how this can be done while still using all the existing dtml-in machinery (ot the TAL stuff for that matter). Any I making any sense here? Just wondering if anyone had come across the same problem.

1 0

Can edit via FTP respect Zope versions?
by Craeg K. Strong 14 Nov '01

14 Nov '01

I would like to see the following scenario: 1) I go into Zope/manage_main, create a Version, and enter it. 2) Outside Zope, I use emacs and pull up a DTMLDocument in an edit buffer using FTP 3) I make my changes and save the file (FTP PUT) 4) I return to Zope and see that the changes have been made INSIDE MY VERSION 5) Happy with the changes, I commit my Version and exit it. Today, this does not seem to work. That is, even though I have started working in a Version, FTP does not respect this. How could such a facility be implemented? Well, what about having a context mapping feature where you could map a user to a context. The Zope FTP machinery, when it gets a login, uses the passed in userid and looks up the context in the (global) context map. For me, it says "oh, Craeg is working in a Version. I will silently and automatically map URLs for GET and PUT to respect his Version." For my colleague, it says "oh, Chip has no special context, no filtering needed..." Further, there could be hooks provided so that Zope developers could provide their own URI Resolution algorithms, so that different user's requests could be mapped in different ways. To use a contrived example, this could be a way of implementing a development workflow scenario-- for development, requests are mapped to one area, where as for testing they are mapped to a different area. Note that the "virtual host monster" operates on a somewhat similar idea. Has anyone else felt the need for such a facility ("URIResolver") ? Does anyone have any pointers where I should look in the Zope libraries to program my own? Thanks in advance! --Craeg

2 1

Zope and python namespaces
by Roch'e Compaan 14 Nov '01

14 Nov '01

I wrote an external method that uses Graphite, Sping and PIL to generate charts, but I ran into some namespace problems. Execution of the external method fails because there is a clash in the namespace between PIL and Zope. PIL has a module "ImageFile" and Zope does as well. Even if I insert PIL's path explicitly as the first entry in the path, Zope's ImageFile is still being used. How can I resolve this and what is the best strategy to avoid namespace problems here? Roché

2 3

CatalogPathAwareness, CatalogAwareness and ZClass CatalogAware (2.3.3)
by Didier Georgieff 14 Nov '01

14 Nov '01

Hello, On 2.3.3 i just discovered a problem with CatalogAware ZClass and VHM. All CatalogAware auto-indexing is fine when i dont use VHM, and when i have (thru Apache and VHM) my root as a folder the autocataloging get the wrong physical path : Example for the same object : www.mysite.com/folder1/item1 is well autoZCataloged as /folder1/item1 (no VHM) but myvirtualsite.com/item1 is missAutoZcataloged as /item1 (with VHM) If i try to "find"/Catalog the objects they are **well** cataloged (with WHM and without VHM). After looking at the list, i found few informations about VirtualHostFix but normally it's fixed and i do think the problem is elsewhere. So : in ZCatalog/__init__.py there is registring for the old CatalogAwareness instead of the new CatalogPathAwareness (with the getURL, getPath wich i use without anu problem in my application) import ZCatalog, Catalog, CatalogAwareness, Vocabulary, ZClasses createZClassForBase( CatalogAwareness.CatalogAware, globals() , 'CatalogAwareBase', 'CatalogAware' ) My questions are : ? Why is there no reference to the new CatalogPathAwareness for CatalogAware ZClass ? I suspect this is the reason why CatalogAware fails and "find" works fine on CatalogAware instance. ? Is this a bug or done by purpose ? ? What happens if i just change the ZCatalog/__init__ with import ZCatalog, Catalog, CatalogPathAwareness, Vocabulary, ZClasses createZClassForBase( CatalogPathAwareness.CatalogAware, globals() , 'CatalogAwareBase', 'CatalogAware' ) Will the thunder will strike me ?? Any help or tip would be more than welcome. Thanks in advance. -- Didier Georgieff Chef de projet technique SIT du Bas-Rhin tél : 03.88.76.78.35 - fax : 03.88.76.78.34 email : didier.georgieff(a)agriculture.gouv.fr

1 0

Signal 11 restarts
by Daniel Duclos 14 Nov '01

14 Nov '01

Hello! I am a Zope newbie, so maybe I am saying some stupid things, but I'm gonna try =). Oh, and forgive my english! I Have a debian linux box running Zope 2.4.0 binary release. I installed MySQLdb-0.9.0 compiled against a Zope 2.1. I am getting signals 11 frequently. I searched the list archives and found this message: http://lists.zope.org/pipermail/zope-dev/2001-August/012878.html But, since I am using a binary release, how can I turn this pymalloc thing off, besides from installing Zope from the source? My system has an old python version <1.5> from debian woody. I can get python 2.1.1 from the source and compile it without the pymalloc, but choose not to install it on the system, and them put this python in the place of zope/bin/python. Will it work? Oh, BTW, i am not getting any cores, despite setting ulimit -c to unlimited <both in the shell and in the start script>, giving wrtie permissions to nobody in the zope dir. Oh, I forgot to mention, my Data.fs has 1 GB of size! Thanks on any advice and for your patiente daniel lobato duclos -- daniduc(a)hiper.com.br -- http://www.hiperlogica.com.br -----------------------------------------------------------------------------

1 0

StructuredText and ZPT nocall() for URL/IMG - suggestion
by Brad Clements 14 Nov '01

14 Nov '01

Just thought I'd mention what I've done with my new website. I'm trying, as much as possible, to make all content be based on Structured Text. I've created a ZClass that combines PageTemplates and stx_to_html.py together. one thing that really annoys me is that the "heres my url":xyz and "my picture":img:wow tag processing in Structured Text assumes that the "urls" are http type paths or relative. I'd prefer that images output tag() values and that urls, when "not local", output absolute_url() values. What I did was pass nocall() to stx_to_html from within my PageTemplate and modified the HTMLWithImages class, as follows in the code below. It tries to use nocall() to lookup the href part of the tag. If an object is found, we use tag() or absolute_url() respectively. Otherwise it outputs the href tag as-is. Currently this is a bit of a hack, but I'm curious if anyone else would like this functionality (in which case it's back to the Wiki). (code fragment from stx_to_html.py) class HWIClass( HTMLWithImages ): def document(self, doc, level, output): # overrides HTMLWithImages.document (Zope 2.4.1b1) to leave out <html> and <body> tag when rendering for c in doc.getChildNodes(): getattr(self, self.element_types[c.getNodeName()])(c, level, output) def image(self, doc, level, output): if hasattr(doc, 'key'): output('<a name="%s"></a>\n' % doc.key) if hasattr(self,'nocall'): try: obj = self.nocall("here/"+doc.href) alt = str(doc.getNodeValue()) alt = alt.replace('"','') output(obj.tag(alt=alt)) hr = None except AttributeError: hr = doc.href else: hr = doc.href if hr: output('<img src="%s" alt="%s">\n' % (hr, doc.getNodeValue())) if doc.getNodeValue() and hasattr(doc, 'key'): output('Figure %s %s\n' % (doc.key, doc.getNodeValue())) def image(self, doc, level, output): if hasattr(doc, 'key'): output('<a name="%s"></a>\n' % doc.key) if hasattr(self,'nocall'): try: obj = self.nocall("here/"+doc.href) alt = str(doc.getNodeValue()) alt = alt.replace('"','') output(obj.tag(alt=alt)) return except AttributeError: hr = doc.href else: hr = doc.href output('<img src="%s" alt="%s">\n' % (hr, doc.getNodeValue())) if doc.getNodeValue() and hasattr(doc, 'key'): output('Figure %s %s\n' % (doc.key, doc.getNodeValue())) def link(self, doc, level, output): """override to use nocall""" if hasattr(self,'nocall'): try: obj = self.nocall("here/"+doc.href) hr = obj.absolute_url() except AttributeError: hr = doc.href else: hr = doc.href output('<a href="%s">' % hr) for c in doc.getChildNodes(): getattr(self, self.element_types[c.getNodeName()])(c, level, output) output('</a>') def stx_to_html(src=None,text="", nocall=None,level=1 ): if src and src.meta_type == 'Rendered STX' and not text: text = src.document_src() st = StructuredText.Basic( text ) if not st: return "" doc = DWIClass( st ) HWICls = HWIClass() HWICls.nocall = nocall html = HWICls( doc, level ) return html Brad Clements, bkc(a)murkworks.com (315)268-1000 http://www.murkworks.com (315)268-9812 Fax netmeeting: ils://ils.murkworks.com AOL-IM: BKClements

1 0

RE: [Zope-dev] Repost to zope-dev: Best way to do "links"
by Bjorn Stabell 14 Nov '01

14 Nov '01

Two comments. One with regards to solving the original problem, one with regards to links. SEVERAL TEMPLATES FOR ONE CONTENT TYPE With the CMF you have another level of typing of objects available: portal_type (dynamic type). A meta_type can have many corresponding portal_types, e.g., a generic MyDocument class (implemented in a Python product) can support many logical types who only varies in their usage and skin. The CMF lets you set the portal_type at creation time (you're basically creating a portal_type object, implicitly based on some kind of meta_type object). What I'm doing now, and it works quite nicely, is to enable users to change the portal_type at run-time. In the Metadata edit view for an object, I scan the available portal_types for the given meta_type and give the user the option to select one of them as the "current dynamic type (portal_type). The CMF's portal_skin tool does the rest of selecting the correct skins to use (based on the portal_type attribute). LINKS IN THE OODB I've used Shane's suggestion, and it works most of the time, but if you ZSync your objects to another server, you'll get two copies instead of one object and one link... Have a look at the PortalHole product release recently; it addresses linking. Another way to do links involves storing the path to the other object (which is how PortalHole does it), or you can create a portal-wide unqiue id (someone suggested this) which you can index in the portal_catalog. Now all you have to do is store the link, and the link won't break if the objects move with respect to each other (=the path changes). An improvement on this is to create a Mixin that implements logic to setup and break these links; this way to you can notify the other object when you're deleted that it should also remove the link. Otherwise you'll have broken links when the other object is deleted. You can also just allow broken links and fix them with regular sweeps across all content objects checking for and removing broken links (sort-of like garbage collection). There should really be a FAQ about how to do this. We've got this kind of Mixin already, but it comes with a few bugs. Let me know if you want/need it. :) Regards, -- Bjorn > -----Original Message----- > From: Oliver Bleutgen [mailto:myzope@gmx.net] > Posted At: Wednesday, November 14, 2001 03:54 > Posted To: Zope Developer > Conversation: [Zope-dev] Repost to zope-dev: Best way to do "links" > Subject: [Zope-dev] Repost to zope-dev: Best way to do "links" > > > Hi reposting to zope-dev because the zope-list didn't yield > any answer (although it should belong there, I think). > > I am unsure how to achieve the following in a product: > > I have a folder with templates which shall be used to render articles. > This folder will be the central repository of templates for > all articles > which find it in their acquisition path. > Now, I want be able to dynamically assign (and change) templates for a > given article, and if I edit one template it should reflect in all > articles which are configured to use that template. > > In http://lists.zope.org/pipermail/zope-dev/2001-May/011187.html Shane > Hathaway describe hardlinks, which seem to do what I need. > > i.e. I then just do > t = template_folder.one_template > self.template = t > > I looked at how ZSQL-methods solve that problem (usage of > SQLConnections), and as far as I can see ZSQL-methods just > store the id > of the ZSQL connection and use that everytime they need to > access the DB. > > Is that right and if so, why is it done that way? > Doesn't give that a performance hit? > > And the last one: > > Will the above described method still allow import/export of these > instances? > > cheers, > oliver > > > > > > _______________________________________________ > 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 )

1 0

Re: [Zope-dev] Catalog Text Indexing
by Lennart Regebro 14 Nov '01

14 Nov '01

From: "Terry Kerr" <terry(a)bizarsoftware.com.au> > Is there any particular reason that a Text Index will index an object by > acquiring the value to index? Shouldn't it not index at all if the > value doesn't exist on the object? If the value is aquired, then it means that this object has that value, aquired. I find it logical that the indexing will behave like that too, and I'm pretty sure I have used it in some application. > I believe KeyWord Indexes have the same problem Why is it a problem? It sounds like a feature to me. :-) If you have an attribute that you don't want to be aquired, I guess you have to make a computed attribute out of it.

1 0

RE: [Zope-dev] Catalog Text Indexing
by Bjorn Stabell 14 Nov '01

14 Nov '01

I find it quite handy that it acquires the value. I often define "global" methods that I call in an acquired way. For instance one could be "MainTitle" which returns a part of the Title only (used to make heads and subheads out of one Title). > -----Original Message----- > From: Terry Kerr [mailto:terry@bizarsoftware.com.au] > Posted At: Wednesday, November 14, 2001 15:04 > Posted To: Zope Developer > Conversation: [Zope-dev] Catalog Text Indexing > Subject: [Zope-dev] Catalog Text Indexing > > > Hi, > > Is there any particular reason that a Text Index will index > an object by > acquiring the value to index? Shouldn't it not index at all if the > value doesn't exist on the object? > > The only way I can get my objects to not be indexed if they don't have > the indexing attribute is to set the attribute to None, reindex, then > remove the attribute. By make this simple change to > lib/python/Products/PluginIndexes/TextIndex/TextIndex.py > > > 353c353,354 > < source = getattr(obj, self.id) > --- > > from Acquisition import aq_base > > source = getattr(aq_base(obj), self.id) > > The object is only indexed if the value exists on that object. This > fixes my problem. > > I believe KeyWord Indexes have the same problem > > Terry > > > > > -- > Terry Kerr (terry(a)bizarsoftware.com.au) > Chief Technical Officer > Bizar Software Pty Ltd (www.bizarsoftware.com.au) > Phone: +61 3 9563 4461 > Fax: +61 3 9563 3856 > ICQ: 79303381 > > _______________________________________________ > 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 ) >

1 0

Catalog Text Indexing
by Terry Kerr 14 Nov '01

14 Nov '01

Hi, Is there any particular reason that a Text Index will index an object by acquiring the value to index? Shouldn't it not index at all if the value doesn't exist on the object? The only way I can get my objects to not be indexed if they don't have the indexing attribute is to set the attribute to None, reindex, then remove the attribute. By make this simple change to lib/python/Products/PluginIndexes/TextIndex/TextIndex.py 353c353,354 < source = getattr(obj, self.id) --- > from Acquisition import aq_base > source = getattr(aq_base(obj), self.id) The object is only indexed if the value exists on that object. This fixes my problem. I believe KeyWord Indexes have the same problem Terry -- Terry Kerr (terry(a)bizarsoftware.com.au) Chief Technical Officer Bizar Software Pty Ltd (www.bizarsoftware.com.au) Phone: +61 3 9563 4461 Fax: +61 3 9563 3856 ICQ: 79303381

1 0