March 2002 - Zope-Dev

RFC : AnyScripts
by Jerome Alet 10 Mar '02

10 Mar '02

Hi, I'm finally toying, after having stalled a too long time, with the creation of ZShellScripts, which would be for ZShell commands what PythonScripts are for Python... The current, not yet available, incarnation is very highly based on PythonScripts, however I wondered if something better couldn't be done, because it seems to me to be an unneeded repetition of most of the PythonScripts' code : AnyScripts AnyScripts would have a mandatory first line similar to the one of any Unix shell script. By unix shell script I mean any script which can be launched directly under Unix, not only bash or csh scripts, but also perl, tcl, etc... So an AnyScript would contain a first line like : #! /scripts/interpreter And the execution of this script in the context on which it is applied would be delegated to another object present in the path which was set on this very first line, i.e. /scripts/interpreter So you could put any interpreter object of your choice in for example the /scripts folder (no path would mean search the interpreter object within the acquisition path) For example PythonScripts would contain something like : #! /builtins/python or something like that. Out of my mind I can already think of a ZShell interpreter object, but why not other types of interpreter objects : perl tcl C (the language ! Why not after all) etc... Each interpreter object would have at least two methods : compile() => returns a code object from source code, if meaningful (otherwise returns the source unchanged), to be stored in the AnyScript object which calls the compile() method. execute() => executes the above code object. This method of the Interpreter object would be called by the AnyScript object at run time, passing it the code object obtained with compile() Of course all of this come straight out of my mind, and there would probably be some problems I'm not able to understand yet... Bound variables would be like for PythonScripts, with the exception of their content would be the best format for the Interpreter in question, so each Interpreter object would probably also have a bind() method which would bind vars in the preferred format for the AnyScript object which calls this method, and would be accessible using the preferred syntax of the langage in question, e.g. : for PythonScripts, vars like context would be Zope Objects, like they currently are. for ZShell AnyScripts, vars like context, would be paths to Zope Objects, and accessible with names like $context for example. for C AnyScripts, vars would probably be something like PyObjects pointers. etc... So AnyScripts objects would interact with their interpreter object of choice to compile their source code, bind their vars, then finally execute the code and get the result back. What do you all think about this idea ? Are any of you interested ? I admit I don't know anything about the feasability of such a project with other languages than ZShell, however if a single implementation could be done, then anyone could add his own interpreter later... PS : I'll do ZShellScripts anyway... bye, Jerome Alet

2 1

Error removing NT Service
by Dean＠work (E-mail) 10 Mar '02

10 Mar '02

Hi I have an application built with Zope which can install, control and remove NT services. The actual code is in an external python script. The services can be installed and controlled but if you try and remove the service it get marked as 'disabled' and can not be reinstalled or remove manually (without using the registry). However if Zope is restarted the service is then removed. Does anyone have any ideas about resolving this problem ?? thanks in advance Dean Houghton

3 2

Usage of try: .. except: .. in zope
by Ivo van der Wijk 10 Mar '02

10 Mar '02

Hi All, I'm currently implementing a QuotaFolder for Zope, i.e. a Folder that restricts the number of objects, their sizes and their total size (and in the future: the installabe meta types). Things look very promising (if anyone feels like testing, please mail me), but it took me quite a while to fix one bug: When creating a ZopeTutorial in a QuotaFolder, I always got Error Type: Bad Request Error Value: The id "examples" is invalid--it is already in use. I couldn't really figure out why my QuotaFolder would cause ZopeTutorial to be installed (or actually imported) twice. Until I studied its code: In addTutorial in TutorialTopic.py, there's the following piece of importing code: # work around old Zope bug in importing try: folder.manage_importObject(tutorialExamplesFile) except: folder._p_jar=self.Destination()._p_jar folder.manage_importObject(tutorialExamplesFile) Basically, anything can go wrong in the first manage_importObject (and in my case it was the raising of a QuotaExceeded exception), causing the code to try to import the file *again* Isn't there a better/cleaner way to do this? It this Exception-type-less try/except construct used elsewhere in Zope (I think I've seen it before)? Should I put a request in the collector? Cheers, Ivo -- Drs. I.R. van der Wijk -=- Brouwersgracht 132 Amaze Internet Services V.O.F. 1013 HA Amsterdam, NL -=- Tel: +31-20-4688336 Linux/Web/Zope/SQL/MMBase Fax: +31-20-4688337 Network Solutions Web: http://www.amaze.nl/ Consultancy Email: ivo(a)amaze.nl -=-

2 1

SQL compare to form value -HELP-
by Sean Kelley 10 Mar '02

10 Mar '02

I am trying to test whether or not 3 SQL fields are equal to a form field. I am doing this within an SQL method. (I am trying to learn zope and Mysql at the sam time) How does one do this? Here is where I am going with this so far: I have these listed as arguments: Breeds Breed1 Breed2 Breed3 SELECT * from Ranch WHERE Breeds = <dtml-sqltest name="Breed1" type="string"> OR Breeds = <dtml-sqltest name="Breed2" type="string"> or Breeds = <dtml-sqltest name="Breed3" type="string"> Breeds is the name of the form input and I want to test and see if it equals any one of the 3 fields Breed1, Breed2, or Breed3 and return all of the row if there is a match. -- Sean

1 0

Trapping Interbase Error Codes
by Beverly Dolor 09 Mar '02

09 Mar '02

Hi! I'm currently doing a program on Zope 2.5 using gvibDa as my database. I would like my program to trap Interbase error codes such as (isc_deadlock, isc_lock_conflict), the problem is I'm still at a lost on how to do this. I saw some progam written in C that trap these kinds of Interbase error using isc_print_sqlerror and isc_sql_interprete, in fact I saw these functions being used on gvibBase.c :) But since I don't know C, the programs didn't help much in finding out how the errors are trapped. I know gvibDa has its own exception handler (gvibExceptions?), but this handler converts the original database error into its own error message (DatabaseError,InternalError, etc). what I want to do is to trap the "raw" database error before it gets converted into one of the customized error messages. And another question, is there a way to commit/rollback a database transaction(using a gvib object) and zope transaction without using an external method? Any answer, opinion, view regarding these issues will be very much appreciated. Thanks :) __________________________________________________ Do You Yahoo!? Try FREE Yahoo! Mail - the world's greatest free email! http://mail.yahoo.com/

3 3

Manual object-to-relational persistency framework
by Bjorn Stabell 09 Mar '02

09 Mar '02

Hello, I'm trying to use Zope without storing everything in ZODB while still having the perception of an object-oriented database. I want to control the object-to-relational mapping layer, but would like the loading and saving of objects to be automatic. Is there some way to do that using Zope? Where would I, e.g., hook into to get automatic loading/cacheing/saving of objects that weren't stored in ZODB, and where I could actually create the methods to load/save objects myself (I want to control the data structure). Regards, -- Bjorn

5 4

zope transaction bug?
by Ross J. Reedstrom 09 Mar '02

09 Mar '02

Hey all - Boy, it's been a long time since I've hung around here. Anyway, I think we've found a bug somewhere in the zope transaction machinery. Every now and then, a backend will get into 'idle in transaction' state. After that, no work on that backend ever commits - the 'END' query just never gets sent. Careful analysis of the log files indicated that the ZpsycopgDA instance started 'losing' proper transactional control after an admin who was logged in via an exUserFolder that _also_ used that DA for authentication used the management interface to undo a Zope transaction. The transaction in question was merely an edit on a DTML file. No db interaction beyond the authentication. Either a successful or an unsuccessful attempt to rollback causes the problem. Seems the Zope transactional machinery has 'forgotten' that our connection needs to be told about transaction boundries. We can make this happen at will, with Zope 2.4.3 or 2.5.0, exUserFolder 0.10.7, and ZpsycopgDA 1.0.5. It does require using the usAuth (User Supplied Auth source) part of exUserFolder. Our developer who's somewhat familiar with the guts of exUserFolder thinks that this is probably still a bug in Zope, since all usAuth does is use customized ZSQL methods to change where various strings are stored in the DB. We're seeing similar problems with ZpopyDA, but haven't tested it thoroughly to see if it's the same trigger. I've constructed the test case for this: it's available at: http://wallace.ece.rice.edu/xact-bug.tar.gz That tarball contains an SQL schema script, zexp, and minimal instructions for replicating the problem. Anyone have any clues as to where to look to fix this? We're losing commits to our database, whenever a developer tries to Undo something: not good! Ross -- Ross Reedstrom, Ph.D. reedstrm(a)rice.edu Executive Director phone: 713-348-6166 Gulf Coast Consortium for Bioinformatics fax: 713-348-6182 Rice University MS-39 Houston, TX 77005

2 3

Shallow copies/links/references
by Mario Valente 08 Mar '02

08 Mar '02

Hi: OK everyone, I've managed to get object links to work. As discussed previously by others (re: object references) and asked by myself: I had the need to be able to refer to objects from different points of the folder hierarchy without duplicating those objects. Yes I know that can be done *programatically*. That is not the idea. The idea is to have a folder with differente widgets/components and to be able to copy and paste references to those widgets throughout the folder hierarchy. Of course that the usual copy/paste is not OK, since that *duplicates* the objects and as such further changes to the original object wont be reflected in the copies. With something like this + Ordered Folders you are able to have an index_html that just renders the objects within a folder and you can then drop objects/widgets into the folder to produce a page. What I did: - edited CopySupport.py - copied the manage_pasteObjects method to a manage_pasteMonikers method - commented the #ob=ob._getCopy(self) line (the duplicate object part) - added a "Paste Ref" button to the lib/python/OFS/dtml/main.dtml file that calls the pasteMonikers method """ <input class="form-element" type="submit" name="manage_pasteMonikers:method" value="Paste Ref." /> """ It works. It does what I want/need. When you change a reference, the original also changes. When you change the original, the references change. You can delete either the reference or the original and all is OK. Issues/TODO - find a way to distinguish references from the original (tried to change the meta_type at paste time but no luck). At least the meta_type/icon should change to provide a visual cue. - provide tool(s) to find impact of changing a reference. Provide a link from a reference object to the original. Provide a link from an object to its several references. Comments and experiences are welcome. C U! -- Mario Valente

2 2

TF & ZMI, was: RE: [Zope-dev] Zope 2.6 and OrderedFolders
by sean.upton＠uniontrib.com 08 Mar '02

08 Mar '02

You know, the funny thing is that TF really accomplishes the goals of a ZMI/UI-improvement hack (unclutter a view of a folder to VISUALLY separate the software/content/presentation space) implemented, IMHO, in the wrong place. I would think there is another way that a Folder UI could be uncluttered without the magic of TF, and that is add "translucency" (items are there, but obscured from view) to the UI layer - that is, the ability to VISUALLY unclutter different users' views of objects in a folder via ZMI; views could be stored per user. For example, you could select ZSQL Methods and RDB connections in a folder to be hidden from everybody but the Database programmer; you could keep these and Python Scripts obscured from the view of a content person or presentation designer, etc. Well, they wouldn't be hidden, but visually 'compacted' and relegated to a small drop-down box at the bottom of the page to unclutter it... so the standard tabled object list wouldn't iterate over objectIds() or objectValues(), but instead a higher-level filter function. A complimentary filter function would render the 'hidden' objects as a sequence as well. __ZMI manage_main Now:______________________ ________________________ ===\===\===\===\======== <--- ZMI tabs % :::::::::::::::::::=== TP.NAME..... .SZ. ..LM.. []---------- ---- ------ []---------- ---- ------ []---------- ---- ------ []---------- ---- ------ []---------- ---- ------ [RN][CT][CP][DL][I/E][SA] <--- Buttons for checked __ZMI With Translucency:___________________ ________________________ ===\===\===\===\======== <--- ZMI tabs % :::::::::::::::::::=== TP.NAME..... .SZ. ..LM.. []---------- ---- ------ []---------- ---- ------ []---------- ---- ------ []---------- ---- ------ []---------- ---- ------ [RN][CT][][][][][Hide] <--- New HideBtn ________________________ Hidden Items: [Select Item... [v] [Open][Unhide][ViewAll] <--- Buttons for doing things ^ with hidden | items | +-----------> ViewAll => pop-up new window ____________________________________________ This would be more visually convenient. I'm thinking of trying to monkey-patch OFS.Folder to just do this. Of course, with these kind of modifications, for any given folder, there's no mandatate that anything need to be hidden at all. Thoughts? Sean -----Original Message----- From: Casey Duncan [mailto:c.duncan@nlada.org] Sent: Wednesday, March 06, 2002 12:02 PM To: sean.upton(a)uniontrib.com; Eric.Roby(a)noaa.gov; zope-dev(a)zope.org Subject: Re: [Zope-dev] Zope 2.6 and OrderedFolders On Wednesday 06 March 2002 02:49 pm, sean.upton(a)uniontrib.com allegedly wrote: > Is this a suggestion that TF go into the core? One other (major) problem > with TF is that it breaks session tracking site-wide, or at least the > ability to have more than one session id manager per Zope instance in CST; > I'm not sure about sessions in Zope 2.5... > > TF seems to have too much normalcy-hijacking magic to go into the core, if > you ask me... > > Sean -1 on TF in the core. Nice hack, but too much magic and there are significant performance implications if I remember correctly. Also Zope 3 will render TF superfluous. /---------------------------------------------------\ Casey Duncan, Sr. Web Developer National Legal Aid and Defender Association c.duncan(a)nlada.org \---------------------------------------------------/

4 3

RE: [Zope-dev] More Zope 2.6 requests
by sean.upton＠uniontrib.com 08 Mar '02

08 Mar '02

+1 on #2 - definitely useful. Sean -----Original Message----- From: Jim Washington [mailto:jwashin@vt.edu] Sent: Friday, March 08, 2002 5:26 AM To: zope-dev(a)zope.org Subject: [Zope-dev] More Zope 2.6 requests While we are asking, I have two things that I would consider valuable additions to Zope 2.6: 1. UserSniffer. Currently an External Method, but has functionality that should be available OOTB to assist making those (horrors!) browser-dependent hacks. It has other uses, too, like explaining to clients that their browser is five years old and needs the free upgrade that is available. 2. Support for HTTP_X_FORWARDED_FOR. Many (most?) of us run zope behind Apache ProxyPass or Squid, and the Zope logs therefore save the ip address of the proxying machine rather than any semblance of where the client browser is. I realize that other proxies do not always follow the rules, etc., etc., but I think that using HTTP_X_FORWARDED_FOR if not null would be better than a log full of 192.168.1.1. Apologies if these have already been discussed under another heading. -- Jim Washington _______________________________________________ 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