- Zope-Dev - Zope lists

RE: [Zope-dev] Weird ConflictErrors on startup
by Toby Dickenson 07 Aug '01

07 Aug '01

> Toby Dickenson wrote: > > > > I think Ive seen this before, when Product initialization modifies a > > persistent object, then causes an exception (ImportError, or > > whatever). You see these ConflictErrors if a subsequent Product > > initialization also modifies a persistent object. > > Ah, okay... > > So, how do you find out what the error causing the problem > is? (the ImportError > in your example) > > The only things showing up in the stupid log file are the > ConflictErrors :-( Ive only seen it after a previous product-initialization failure traceback. sorry....

2 1

Weird ConflictErrors on startup
by Chris Withers 07 Aug '01

07 Aug '01

Hi, I'm trying to move a project from Zope 2.3.2 to Zope 2.4.0. Once I move to 2.4.0, I get lots of the following errors on startup: 2001-08-06T11:32:26 ERROR(200) Zope Couldn't install Membership Traceback (innermost last): File /usr/local/zope/2.4.0_base/lib/python/OFS/Application.py, line 713, in install_product (Object: ApplicationDefaultPermissions) (Info: Membership) File /usr/local/zope/2.4.0_base/lib/python/ZODB/Transaction.py, line 302, in commit File /usr/local/zope/2.4.0_base/lib/python/ZODB/Connection.py, line 420, in commit (Info: (('App.Product', 'Product'), '\x00\x00\x00\x00\x00\x85\x0e\xb5', '')) File /usr/local/zope/2.4.0_base/lib/python/ZODB/FileStorage.py, line 663, in store (Object: dev/var/Data.fs) ConflictError: ('\x00\x00\x00\x00\x00\x00\x00\x00', '\x03>9\xad\xbf;\xcbD') The complete list of these errors, in order, is: LoginManager Membership PlugIns PythonScripts StandardCacheManagers ZPatterns ...they all report the same identical ConflictError, including the OID-y bits. I'd love to know what's going on here but I'm completely stumped... cheers, Chris

3 5

ANNOUNCE: phabric project - Zope on Java
by Kevin Dangoor 07 Aug '01

07 Aug '01

A few weeks back, Web Elite started work on Phabric, a project to bring Zope to Java via Jython. Since there has been a lot of discussion in the past couple of weeks on the jython-dev list about Zope related topics (ExtensionClass, Acquisition and Zope as a whole), we've decided to fully open the project now rather than telling interested people individually. Web Elite is committed to this technology. This open source Java software will bring many of the benefits to Java developers that Python developers have been enjoying already. The main phabric website is at http://www.phabric.org. If you're interested in phabric, we have started some mailing lists to get some discussion going. You can sign up for the lists at http://lists.webelite.com. If you're just interested in tracking the general progress of the project, subscribe to phabric-announce. For developers interested in working on phabric, we have set up a phabric-dev list to discuss the core phabric work. Current Status -------------- ExtensionClass and Acquisition are fairly complete. We're working on the other C modules (largely in python to begin with) right now. The next step after that will be to do profiling and move some of the code into better-optimized Java. The latest download information for phabric will always be at http://www.phabric.org/download. Currently, the only access to phabric is via CVS. Why phabric? ------------ We chose to call the project phabric because there will not be 100% compatibility between zope and phabric, and a new name will avoid confusion. "Phabric" was chosen because it's a framework that holds your web apps together. Metaphorically speaking, you can weave your objects into the phabric of the web... I am sending this message to jython-dev and zope-dev, but not crossposting to avoid discussions that are irrelevant to the other group. Hope to hear from you soon! Kevin Dangoor Web Elite

1 0

Java Python Extension and Zope
by Tim Hoffman 07 Aug '01

07 Aug '01

Hi I have just started experimenting with the JPE (Java Python Extension - http://www.python9.org/p9-cdrom/10/index.htm & http://www.arakne.com/jpe.htm ) as presented at Python9. I can successfully access java classes/methods through external python methods called from dtml (products come next ;-). (only just got it to build and run, with only the most simple of tests, now I need to get my head around the issues of returned data types, and what it will mean to persistance ;-) I think this technology holds all sort's of interesting possibilities to extend Zope so that it could make native rmi calls to java services, maybe you could use JSP's as backend components (a bit like the PHP adaptor, mind you one might ask why ;-). If nothing else it can go some way to alleviating some of the regular objections I have encountered promoting zope, that talking to java services through mechanisms other than xml-rpc isn't really feasible. (one thing that interests me is taking some standard java implementations of services and bolting them on to zope ie JXTA (mind you there may well already be a python implementation in the works ;-) Is anyone else exploring this stuff? If so drop me a line. If any one else is interested I'll put up a wiki page to start a discussion going (benefit's, technical issues etc...) Rgds Tim -- Tim Hoffman Zute Pty Ltd mobile: 0411 068888 fax: +61 8 6210 1883 email: timh(a)zute.net

1 0

changing aquired attributes
by Matt 07 Aug '01

07 Aug '01

Hopefully a simple answer to this. I am using an external method to change aquired properties, eg, to increment an integer property acquired from somewhere. The problem is to persist this. It is easy enough to find the aquired property, eg: self.aquired_property, and then of course to change it. But to make this persist, the only way I found of doing this was to write a recursive function that went through each aquisition wrapper asking if this unwrapped level had the property. If it did I would set self._p_changed=1 on the object that owned it and all would be fine, the update to the property would persist. Is there any more elegane method for updating an aquired property like this that also forces it to persist? regards Matt

1 0

Memory Leak
by J�lio Dinis Silva 07 Aug '01

07 Aug '01

Hi, I'm running zope232src on Linux. My Data.fs packed has a size near 250 MB. My zope server takes 70Mb of RAM when started, ending after a week taking 1GB of RAM, and eventually restarts automatically (z2.py -Z). When it is at 1GB of RAM there is on top of refcounts in Control_Panel/DebugInfo the class BTrees.IIBTree.IIBTree with something like 3500 refcounts. Then there is OFS.Image.Image with 1000 refcounts and other classes with refcounts below this. I have a lot os catalogs in my Data.fs some with 10,000 others with 5,000 objects indexed. I've not converted all of my zcatalogs to the new Btree implementation with manage_convertBtrees. But now I'm thinking of doing that since I think the memory leak is in Btrees. I also have a BtreeFolder from the product BtreeFolder with 15000 folders created in it, but I dont think it could be from here. I'm doing debug to find this memory leak. Any hints if this memory leak could really be related with Btrees? Best Regards, JS _________________________________________________________________ Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp

1 0

ZOracleDA
by Juan Javier Carrera Obrero 07 Aug '01

07 Aug '01

Hi, I am trying to install Zope in Windows 95 or Windows 98. It runs ok. However, I have installed the ZOracleDA product, but this product does not run correctly. This problem is caused because the library "buffer.dll" and "oci_.dll" are not enabled for Windows 9x. I have been compiled and trying to link the sources (in the "src" directory), but how the Makefiles and others files are prepared for Unix or Linux, I can not do it. Anybody knows to how to do it?, or anybody can send me these "dlls" for Windows 9x? Thank you very much.

2 1

In-memory persistence of non-serializable objects?
by Michael Strasser 06 Aug '01

06 Aug '01

I want to create a Zope object that can keep non-serializable objects (specifically sockets) in memory between requests. The idea is (in its simplest form): * On first request to the object a socket is opened and stored. * On subsequent requests the same socket is used. If one of these requests is to close the socket it is closed. If the socket has been closed by the other end an error is returned. I haven't found information about in-memory persistence of Zope objects and non-serializable objects can't be stored in ZODB. Have I missed something obvious? Thanks in advance. -- Michael Strasser Brisbane, Australia

3 2

Re: [Zope] can one pass values through the 'action' property?
by Florent Guillaume 06 Aug '01

06 Aug '01

[followups to zope-dev] Tommy, You're right, I did the tests and it is indeed a Zope problem. What happens is this. If you have: <form action="/action?row=3" method="POST"> <input type="submit" name="submit" value="submit"> </form> Then what gets sent by Mozilla to the web server is: POST /action?row=3 HTTP/1.1 Host: ... ... Content-Length: 13 submit=submit And indeed there is enough information here to build a correct REQUEST with both row=3 and submit=submit. (However note that with a GET method, Mozilla sends: GET /action?submit=submit HTTP/1.1 So the GET method wouldn't work here anyway.) The culprit is ZPublisher.cgi.FieldStorage, who only looks at the QUERY_STRING if it's a GET or a HEAD. If everybody else does it, maybe Zope should do it... A word about the spec: HTML-4.01 says: * If the method is "get" and the action is an HTTP URI, the user agent takes the value of action, appends a `?' to it, then appends the form data set, encoded using the "application/x-www-form-urlencoded" content type. The user agent then traverses the link to this URI. In this scenario, form data are restricted to ASCII codes. * If the method is "post" and the action is an HTTP URI, the user agent conducts an HTTP "post" transaction using the value of the action attribute and a message created according to the content type specified by the enctype attribute. So the "get" part isn't even clear, what happens if the action already contains a '?'... the spec doesn't say. Note BTW that an URI with two '?' wouldn't be legal. Cheers, Florent Guillaume Nuxeo > Florent, > > Ok, if it works in Coldfusion and ASP, then there isn't a problem with > 1. My browser > 2. HTTP spec > 3. Development language > > But if, now, I try to do the exact same thing in Zope, and it doesn't work, > then how could you NOT say that it isn't a Zope problem? I'm not totally > assuming that Zope is the problem, (assumptions can be deadly), but if it > has worked before, and doesn't work now with a new development language, > then where is the first place one should look? > > I'd be interested in input from developers who may have done this in PHP or > other languages. If it also works in them.... > > The only thing that I can think of is if Zope parses differently from other > languages. Meaning that if you choose to pass variables through the action > property, then you MUST use Get rather than Post. In the other languages it > didn't matter, but with Zope, it could potentially matter. I'll test it and > write back.

1 0

LDAP and local roles
by Florent Guillaume 06 Aug '01

06 Aug '01

I have a problem which can easily be solved by local roles: - several workgroups (each one is basically a folder) - for each workgroup, a list of users who are allowed to do stuff in it What I do is create a role WorkgroupMember, and in each workgroup assign a local role of WorkgroupMember to the users who are authorized. However I'd like all this information about workgroup membership to be stored in an LDAP directory, where it belongs. Currently I'm a bit stuck. What I envision would be something akin to LDAPLoginAdapter but whose effect would be on the local roles of a given folder. In my LDAP directory, I would have something like: dn: cn=WorkgroupMember, ou=workgroup1, ou=workgroups, dc=example, dc=com objectClass: groupOfUniqueNames uniqueMember: cn=bob, ou=people, dc=example, dc=com uniqueMember: cn=pete, ou=people, dc=example, dc=com uniqueMember: cn=joe, ou=people, dc=example, dc=com And the local roles in the folder would be parametrized by: groups base DN: ou=workgroups, dc=example, dc=com groups attribute: ou local role name attribute: cn login name attribute: cn user base DN: ou=people, dc=intercom, dc=gouv, dc=fr user RDN attribute: cn (all this can be factored somewhere) group: workgroup1 Well, that's how I see it. I have three questions: - is there a better way to do it ? - has someone done something like that ? Would a LDAPified Slave User Folder fit the bill ? (BTW, I should do a separate message about this, but Slave User Folder and LDAPLoginAdapter don't mix well, as LDAPLoginAdapter.getUsers returns only *cached* users... It could work if Slave User Folder only used getUserNames when listing available users) - If i do it with local roles, where and how would I plug the machinery to do this ? I need to replace the RoleManager, but I don't want to re-derive all classes. I could patch the folder to have a callable __ac_local_roles__, and I think things would mostly work (AccessControl.User does the right thing with callable __ac_local_roles__), but I'm not completely sure, beside that's ugly. Thanks, Florent Guillaume Nuxeo

1 0