February 2000 - Zope-Dev

Re: [Zope-dev] ExtensionClass question
by James Henstridge 25 Feb '00

25 Feb '00

On Thu, 24 Feb 2000, Phillip J. Eby wrote: > At 05:33 PM 2/24/00 -0800, Michel Pelletier wrote: > The method chain of an ExtensionClass type structure can point back to a > previous method chain, thus allowing extension of the type. Similarly, the > other members of the type structure can point to existing routines. And, > as long as one uses a backwardly compatible data structure for the type, > one can subclass an existing ExtensionClass. It is usually easier, > however, to just create mixin ExtensionClasses for the different bits you > need, and combine them together as base classes for a Python class. > I had read the stuff about method chains, but just using method chains does not store any relationship between the two classes. If possible, I would like to be able to use subclass relationships for a bit of type safety in my code (ie. only accept ExtensionClasses of type xxx or its subclasses rather than accept any ExtensionClass). It must be possible to do something like this, as subclassing an ExtensionClass in python creates a new ExtensionClass. Doing the same in C must also be possible (although maybe not with ExtensionClass's current public API). James.

1 0

ExtensionClass question
by James Henstridge 25 Feb '00

25 Feb '00

I posted a question to this list about the ExtensionClass module that seems to be maintained by the zope developers now. I haven't received a response yet, so I was wondering if this was the right place to ask, or was it because I posted on the weekend and the message got ignored? Rather than reposing it, I will just put a link to it in the archives: http://lists.zope.org/pipermail/zope-dev/2000-February/003389.html It is on subclassing ExtensionClasses in C. James. -- Email: james(a)daa.com.au WWW: http://www.daa.com.au/~james/

4 6

Question about setting roles
by James W. Howe 24 Feb '00

24 Feb '00

I'm trying to track down why my subclass of folder seems to be having problems with permissions. I've discovered that if I override the definition of manage_main with my own template, security doesn't work correctly for that template. In particular, when the filtered options code is executed, my manage_main only has a role of 'Manager', but my folder instance allows 'Manager' and 'Publisher' for View Management Screens. The one thing I don't understand is how roles are assigned to template objects. For example, where/when were the roles assigned to my manage_main template? I've worked through the code to see how the manage_changePermissions works, but that only seems to set the permissions for the containing object, in this case my Folder subclass instance. In my case, it sets the _View_management_screens_Permission to ['Manager', 'Publisher']. How and when does this information make it back to the objects associated with this permission? How (when/where) would the roles be set for my manage_main template? I'm going to continue to look at things to see if I can figure this out myself, but it would help a great deal if someone could point me in the right direction. Thanks! James W. Howe mailto:jwh@allencreek.com Allen Creek Software, Inc. pgpkey: http://ic.net/~jwh/pgpkey.html Ann Arbor, MI 48103

1 0

RE: Scheduler Transactions (was Private Product Objects)
by Martijn Pieters 24 Feb '00

24 Feb '00

From: Loren Stafford [mailto:lstaffor@dynalogic.com] > Moving over to zope-dev... Good idea. > > Martijn, don't worry. I'm not going to change designs in > mid-stream. But I > still have to think thru all the possibilities, perhaps > covering ground that > you've already explored and left behind. > > From: Martijn Pieters <mj(a)digicool.com> > Sent: February 24, 2000 03:24 AM > Subject: RE: [Zope] Private Product Objects (was: object icons) > > > > (Events are called using ZClient, because the Events have > to be called > > independent of current web requests, and you do need a HTTP REQUEST > > context). > > I assumed from the beginning that the event Dispatcher would > fire events > with RPC (ZClient or something like it) for the reasons you > mention, plus > the fact that the thread for the event is allocated by existing Zope > mechanisms. > > Then I started thinking about transactions and atomicity. You > don't want a > system failure to result either in loss of an event or > duplication of an > event. But how can you ensure that in the case of the Scheduler? > > So, I started wondering how to atomically coordinate the > updating of the > Schedule catalog (to disarm the event) with the execution of > the event's > action method. Why should this be atomic? The integrity of the Catalog is most important. First, in it's own RPC call, update the catalog (disarm or reschedule next call). Then, separate from this, call the Event. Yes, you run the risk of not calling a certain event in case of a system failure. But in that case the event could not have run anyway. I'd say, solve it like I'd do, and move any reliability issues to a next release. We are replacing cron here, not MessageQueue. -- Martijn Pieters, Software Engineer | Digital Creations http://www.digicool.com | Creators of Zope http://www.zope.org | mailto:mj@digicool.com ICQ: 4532236 | PGP: http://wwwkeys.nl.pgp.net:11371/pks/lookup?op=get&search=0xA8A32149 -------------------------------------------

1 0

RE: XMLDocument "WrongDocument" error and object comparision...
by Amos Latteier 24 Feb '00

24 Feb '00

> I'm playing with XML Document, and I'm finding that I'm getting a > bogus WrongDocument error from the sample file. If I open up the > 'slides' document and hit 'Change' I'm getting a WrongDocument > exception at line 433 of Node.py. > > if document and ownerDocument and document != ownerDocument: > raise ZDOM.WrongDocumentException() > > (I also see this when I try to create a new document) > > Putting a > print "DOC", document, ownerDocument > print "DOC2", id(document), id(ownerDocument) > on the line before it gives back > > DOC <Document instance at 863bf90> <Document instance at 863bf90> > DOC2 139736152 139811552 > > so they _look_ like the same object, but comparision is failing. > > On a hunch, I tried changing the code to compare document.aq_base > and ownerDocument.aq_base, and this works. I have _no_ idea > if this is > correct, though. I think that you are right. Let me look at it a little more, and then I'll check in your patch. Thanks! -Amos -- Amos Latteier mailto:amos@digicool.com Digital Creations http://www.digicool.com

1 0

RE: [Zope-dev] XMLDocument - Importing XML and assigning IDs to r esulting nodes
by Amos Latteier 24 Feb '00

24 Feb '00

> How do I import an XML document such that the IDs of the > resulting nodes is > derived from an attribute associated with the source XML tags. > > Ive tried using <tag id="foo"> but I get the following error: > > Error Type: Bad Request > Error Value: Invalid or duplicate property id > > Is there a way to do this, or are the nodes of an imported > XML file destined > to be given cryptic, numbered IDs? Unfortunately right now you cannot control the IDs of the nodes. This is a big problem with the current implementation. We definitely plan to fix this in later implementations. This problem along with a number of others are why XML Document is currently considered 'alpha' quality. -Amos -- Amos Latteier mailto:amos@digicool.com Digital Creations http://www.digicool.com

1 0

Re: [Zope-dev] Functional programming
by Tres Seaver 24 Feb '00

24 Feb '00

Lalo Martins <lalo(a)hackandroll.org> asks: > To: zope-dev(a)zope.org > Subject: > > I know it probably is on the archives, but I didn't find it. > > Is there a good reason why I can't use map, filter and lambda > in DTML code? (They're not in the namespace.) Duncan Booth <duncan(a)rcp.co.uk> replies: > > lambda is available, but you have to specify an extra default > argument _vars=_vars > > Is there a good reason why I can't use map, filter and lambda > > in DTML code? (They're not in the namespace.) > > > Even without map and filter this can be useful on occasion if you > find yourself otherwise repeating an expression a lot of times. On > the other hand it isn't very pleasant to look at. > > I think the answer is probably that if you find you need to use map > and filter then you should consider using Python instead. Besides > you can get the effect of both of these with dtml-in. Lalo volleys: > > > > I think the answer is probably that if you find you need to use map > > and filter then you should consider using Python instead. > > Not really. Practical example: > > <dtml-in "map (Catalog.get_object, Catalog(...))"> > perform some operation, like listing the objects, without > need to bload the Catalog with all properties in the objects > </dtml-in> The downside here is that you activate every object found, which can be deadly for a really large result set. The rationale for using meta_data is to allow presentation of a limited view of the catalogued objects without pulling them into memory from the ZODB. > > And even if I _should_ use Python, then it should at least be > supported in PythonMethods. PythonMethods have no more access than DTML Methods/Documents (by design). > > > Besides you can get the effect of both of these with dtml-in. > > But the code you need for that looks mightly ugly, with > temporary variables, lots of REQUEST.set... let's not go there > :-) Agreed. filter() is particularly nice, as it eliminates the nesting of the <dtml-if> inside the <dtml-in>. For instance, to grab all contained objects which have a "browseable" property, set to true: <dtml-in "_.filter( lambda x: x.hasattr( 'browseable' ) and x.browseabe, objectValues() )"> ... </dtml-in> I will admit that I can't think of a particular use for map() offhand (but then, I'm not a lambda-head, either). -- ========================================================= Tres Seaver tseaver(a)palladion.com 713-523-6582 Palladion Software http://www.palladion.com

3 2

RE: [Zope-dev] Folder Subclassing Problem
by James W. Howe 24 Feb '00

24 Feb '00

At 05:35 AM 2/24/00 -0500, Martijn Pieters wrote: >From: James W. Howe [mailto:jwh@allencreek.com] > >[Synopsis: > Created a subclass of Folder > Defined my own manage_main > Created instance of my folder subclass > Defined a new role for the folder and permitted the role to view > management screens, contents, but nothing else. > Tested for superuser and things worked. > Tested for my custom role and only saw the index_html] > >I still don't know for sure, as I don't have the time to investigate, >but I think there are some problems with the security interface, making >security too tight. It just doesn't work as expected. > >Try this: Create your Role and user in the top-level acl_users folder. >Then give that Role permission to view management screens either on the >top-level folder or on your own product. I tried what you suggested above, but the results were the same. I then decided to comment out the code in my subclass where I define my manage_main (i.e. manage_main = HTMLFile(...)). I reran my tests and this time both my superuser and my custom role were able to see the management interface and the contents screen (aka manage_main). For some reason, when my subclass uses its own manage_main, the permissions don't seem to work properly. Further evidence of this comes from some debugging I did in Management.py. I stepped through the filtered options code. When the "contents" object was asked for it's roles, it answered "Manager". Naturally, when the user wasn't the manager, this option got filtered out because the user didn't have access to it. In my security screen, I have View management screens, and Access contents information checked for both my manager and my custom role. I then modified my code to eliminate my definition of manage_main and instead use the one defined by ObjectManager. I then stepped through the debugger once again. This time, when the roles of the manage_main object were given, it came back as a collection of 'Manager', 'Publisher' (my new role), 'Manager'. I'm still investigating why this might be happening. James W. Howe mailto:jwh@allencreek.com Allen Creek Software, Inc. pgpkey: http://ic.net/~jwh/pgpkey.html Ann Arbor, MI 48103

1 0

Scheduler Transactions (was Private Product Objects)
by Loren Stafford 24 Feb '00

24 Feb '00

Moving over to zope-dev... Martijn, don't worry. I'm not going to change designs in mid-stream. But I still have to think thru all the possibilities, perhaps covering ground that you've already explored and left behind. From: Martijn Pieters <mj(a)digicool.com> Sent: February 24, 2000 03:24 AM Subject: RE: [Zope] Private Product Objects (was: object icons) > (Events are called using ZClient, because the Events have to be called > independent of current web requests, and you do need a HTTP REQUEST > context). I assumed from the beginning that the event Dispatcher would fire events with RPC (ZClient or something like it) for the reasons you mention, plus the fact that the thread for the event is allocated by existing Zope mechanisms. Then I started thinking about transactions and atomicity. You don't want a system failure to result either in loss of an event or duplication of an event. But how can you ensure that in the case of the Scheduler? So, I started wondering how to atomically coordinate the updating of the Schedule catalog (to disarm the event) with the execution of the event's action method. I thought of having the Dispatcher wrap a transaction around the event's transaction, but that doesn't solve the problem. Transactions don't nest well. You could still have a failure between the end of the event's transaction and disarming of the event. I thought of using LPC (direct call) instead of RPC so that the Dispatcher could define the transaction in which the event runs and include the Schedule update in the same transaction. That sound great from the transaction point of view. But then the Dispatcher would have to manage threads for the events (not KISS), plus there would be no HTTP REQUEST context, etc. In short, I don't have a solution. Any ideas? -- Loren

1 0

RE: [Zope-dev] Accessing root object
by Martijn Pieters 24 Feb '00

24 Feb '00

> From: Robert Sander [mailto:zope-dev@beteigeuze.cs.tu-berlin.de] > > On Thu, Feb 10, 2000 at 03:40:38PM -0000, Phil Harris wrote: > > I think there's something called app? > > > > It's declared in <zopedir>/lib/python/Main.py. > > > > That may be what your looking for?? > > I don't know, the object app is not in the global namespace > when __init__.py > of my Product is called. > > It may be the same problem with the discussed scheduler > product when creating > another thread, b ut I don't know. Here's what I'd do (in __init__.py): def initialize(context): app = context._ProductContext__app # Don't you love private name mangling if not hasattr(app, #IdForObject#): object = Object() app._setObject(#IdForObject#, object) get_transaction().note('Added Object') get_transaction().commit() context._ProductContext__app is the root object. #IdForObject# is a string ID for the object. -- Martijn Pieters, Software Engineer | Digital Creations http://www.digicool.com | Creators of Zope http://www.zope.org | mailto:mj@digicool.com ICQ: 4532236 | PGP: http://wwwkeys.nl.pgp.net:11371/pks/lookup?op=get&search=0xA8A32149 -------------------------------------------

6 13