March 2001 - Zope-Dev

LoginUser and Dictionaries
by Johan Carlsson 06 Mar '01

06 Mar '01

Hi, I'm working on a User object that will support Groups and Groups-to-Roles mappings. I modeled it a bit and come up with this model: (I use Coad's constraints notion, putting constraints with the item the constrain) User -n------n- Role | n | | 1 GroupRoleItem -1----- n- Role (*same as the other role above) | 1 | | n Group To make it as simple as possible I whan to keep Group and GroupRoleItem in a dictionary inside the User instance. E.g. {'groupname': ('rolename','rolename',), ... } In my LM implementation I want to store GroupRoleItems in a SQL table and Groups in another SQL table (which records awailable groups). The LoginUser object should have a attribute grouproles_dict that gets its data from the GroupRoleItems table. Here's some code: class GroupLoginUser(LoginUser): ... def getGroupRoles(self): # group_list contains the following brains from the LoginManager (Skinscript/SQL method): # (brains.username) # brains.rolename # brains.groupname # groups = () grouproles = {} if self.group_list: for g in self.group_list: if g.groupname in groups: grouproles[g.groupname].append(g.rolename) else: grouproles[g.groupname] = [g.rolename] groups = groups + (g.groupname,) return grouproles The GroupLoginUser is not stored persitently in my implementation, but I like to cache the grouproles for as long as possible. For example I'm planning a method "getGroups" in the GroupLoginUser which could just return grouproles.keys(). So because GroupLoginUser is a virtual dataskin I'm not sure where I should save the grouproles. In self._v_grouproles or as a skinscripted attribute? E.g. WITH self.getGroupRoles() COMPUTE grouproles = RESULT I might be thinking wrong in the model as well, so any hints and tips for corrections are welcome. Regards, Johan Carlsson

1 0

Class attr hotfix
by Casey Duncan 05 Mar '01

05 Mar '01

I noticed that the most recent hotfix completely denies access to the getClassAttr, setClassAttr and delClassAttr from DTML. Is this going to be the behavior for future Zope versions? I know this will break code, so it might be prudent to alert people when the fix is rolled into the next Zope release. I would like to argue for deprecating these functions in favor of making the ZClass act like a dictionary so that: ZClass.set('name',value) ZClass.get('name'[,inherit]) ZClass['name'] could be called securely from scripts. I would be willing to write the code for this if you think it would be a good idea. ===== | Casey Duncan | Kaivo, Inc. | cduncan(a)kaivo.com `-----------------> __________________________________________________ Do You Yahoo!? Get email at your own domain with Yahoo! Mail. http://personal.mail.yahoo.com/

4 3

is there a hook for before the transaction is committed
by Tim McLaughlin 05 Mar '01

05 Mar '01

Is there a hook for before the transaction is committed for objects which subclass Persistent? I found __inform_commit__ for a "registered" object, but I can't seem to get that to work as I thought it did. I also tried subclassing TM like a DA, but to no avail. TIA, Tim ___________________________________________________________ Tim McLaughlin BCSwebservices.net Director, Technical Group 1950 Old Gallows Road tel: (703) 790.8081 x111 Suite 201 tim(a)bcswebservices.net Vienna, VA 22182 www .bcswebservices. net

4 4

namespace binding in pythonscript
by R. David Murray 05 Mar '01

05 Mar '01

Has the problem with the binding of the namespace in pythonscript been fixed in cvs (if so will it be in b2)? It's a pain to having to call my pythonscript methods using xyz(_=_). --RDM

1 1

RE: [Zope-dev] is there a hook for before the transaction is comm itted
by Tim McLaughlin 05 Mar '01

05 Mar '01

In what respect are the _v_* attribs gonna cause problems. My guestimate was that they disappeared upon transaction commit/abort. I'm also not sure as to why I would need subtrans since I'm only messing with properties of the object. To my knowledge, subtrans are only necessary to conserve resources. Anyway, what you gave me works! Thanks. It seems I need to override the _vote method (since it is only called once and allows exceptions). As to the other stuff, I'm sure you can enlighten me further as to why. I appreciate the help. Cheers. Tim -----Original Message----- From: John D. Heintz [mailto:jheintz@isogen.com] Sent: Monday, March 05, 2001 12:45 PM To: Tim McLaughlin Cc: 'zope-dev(a)zope.org' Subject: Re: [Zope-dev] is there a hook for before the transaction is committed Hi Tim, I have two suggestions, I hope one of them helps. 1) Attached is a TM.py file that I wrote based on the one you mention below. I've tried to make it more obvious and better documented. 2) Don't use this kind of functionality, but rather use sub-transaction commits. The first suggestion has more overhead than what I assume you would need, but the second one won't work for all situations. A Fishbowl proposal of mine, HashingSupport, was going to use the same kind of hook you are asking about. In this case though, using sub-transaction commits made a lot more sense. In general though, I think that _v_* attributes pose a non-trivial problem that probably requires a hook on abort() if not commit() as well. John Tim McLaughlin wrote: > Is there a hook for before the transaction is committed for objects which > subclass Persistent? I found __inform_commit__ for a "registered" object, > but I can't seem to get that to work as I thought it did. I also tried > subclassing TM like a DA, but to no avail. > > TIA, > Tim > > ___________________________________________________________ > Tim McLaughlin BCSwebservices.net > Director, Technical Group 1950 Old Gallows Road > tel: (703) 790.8081 x111 Suite 201 > tim(a)bcswebservices.net Vienna, VA 22182 > www .bcswebservices. net > > > _______________________________________________ > 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 ) -- . . . . . . . . . . . . . . . . . . . . . . . . John D. Heintz | Senior Engineer 1016 La Posada Dr. | Suite 240 | Austin TX 78752 T 512.633.1198 | jheintz(a)isogen.com w w w . d a t a c h a n n e l . c o m

2 1

Zope Performance - Windows 2000 .. Newbie
by Bryan Baszczewski 05 Mar '01

05 Mar '01

I am getting some serious Zope lag time (15-20 seconds) for simple HTML pages when also trying to hit the same port with data requests. The data requests are <= 50 records of text data and SQL Server is optimized (meaning the querys dont take that long, just the connection to Zope). We are using port 8080 on Windows 2000 and was wondering if there is some kind of queue or prioritizing of Zope requests? If so, how can I work around this? Thanks in advance!

1 0

Apology! (was: [Zope] Problems with DTML-IN patch!!!)
by Dieter Maurer 05 Mar '01

05 Mar '01

John Morton writes: > On Fri, 2 Mar 2001 20:55:40 +0100 (CET) Dieter Maurer <dieter(a)handshake.de> > wrote: > > > Chris Withers writes: > > > John Morton wrote: > > > > > > > > It doesn't appear to be defined in DT_In.py in 2.2.5, but it is ment > ion in > > > > it's doc string, and is available in the dtml-in namespace. I never > got > > > > around to tracking down where it was added in. > > "DT_InSV.py" (for DTML In Sequence Variables) defines it, if > > it is defined (when I recently looked through the sources, it has > > not been). > > I can't find them in either DT_In.py or DT_InSV.py in 2.2.5, yet they do > actually work fine... You are right! "{next,previous}-sequence-start-number" and friends are still there. Nobody made the stupidity to remove them. Please (Brian) accept my apology for my quite rude accusation. I was just not smart enough to understand the ingenious code in "DT_In" and "DT_InSV". Let's review the situation: "DT_In" only defines "next-sequence-start-index". But what is pushed onto the DTML namespace is in fact a "DT_InSV.sequence_variables" object. Its "__getitem__" is used to look up sequence variables. The "__getitme__" code is quite sophisticated: If the key is not defined, it splits it at the last '-' into prefix and suffix (in our case, suffix is "number"). If the class "sequence_variables" has a method whose name is given by suffix (this is the case for "number" and many of its friends), then the corresponding "index" key is looked up and the method applied to it. This way, "next-sequence-start-number" and friends come into existence, without that one easily recognized how this happens. Sorry for the false accusations! Dieter

1 0

RE: [Zope-dev] is there a hook for before the transaction is comm itted
by Tim McLaughlin 05 Mar '01

05 Mar '01

I want to be able to raise an event from the property sheets to the product if a property changes. I'm calling it manage_afterChange. I want it only to fire once per transaction so that multiple changes are aggregated. Here's the declaration I have in mind. manage_afterChange(oldItems, newItems) oldItems: dict of id - values before modifications newItems: dict of new values at end of transaction This would allow an elegant "reindex" or notification system for objects. It would also allow for validation (ie. raise error if you don't like newItems). Anyway, that's it. I'm halfway there, just can't get a method to be called on my object _just_ before commit. Cheers, Tim ps. I'm not so sure that __getstate__ will do this for me, but that's probably cause I have no idea what __getstate__ does. It just doesn't sound right ;-) -----Original Message----- From: Toby Dickenson [mailto:tdickenson@devmail.geminidataloggers.co.uk] Sent: Monday, March 05, 2001 12:08 PM To: Tim McLaughlin Cc: 'zope-dev(a)zope.org' Subject: Re: [Zope-dev] is there a hook for before the transaction is committed On Mon, 5 Mar 2001 10:56:38 -0500, Tim McLaughlin <tim(a)BCSWebservices.net> wrote: >Is there a hook for before the transaction is committed for objects which >subclass Persistent? __getstate__ ? But why would you want that? Toby Dickenson tdickenson(a)geminidataloggers.com

2 1

Re: [Zope-dev] DB connection pooling
by Dieter Maurer 05 Mar '01

05 Mar '01

Erik Stephens writes: > I've been working on an implementation of connection pooling for the > MySQLDA. Basically, Zope connections will be pooled by connection > string to reduce the number of unecessary connections to MySQL. Isn't that trivial with Zope (thanks to acquisition)? Just instantiate a single MySQL Database Connection per connect string (sufficiently high above in the site hierarchy). Use it below through acquisition. Dieter

2 1

How is memory shared between Zope threads?
by Erik Stephens 05 Mar '01

05 Mar '01

Hello all, I have noticed incosistencies between Zope threads when objects change. It appears that some threads see the new image of the object while others see the old image. Can anyone enlighten me on how Zope ensures that all threads are seeing the same images? Or at least point me to the modules in the source that handle these matters? Naively, I can think of 2 approaches: a shared memory approach or a separate memory per thread with notifications. Zope behaves like each thread as its own copy of in-memory pieces and sometimes these per/thread memory blocks do not get updated. System info: Zope 2.1.6, Redhat 6.2, PC arch Thanks, Erik

3 4