January 2000 - Zope-Dev

Authentication Patch
by Evan Simpson 12 Jan '00

12 Jan '00

I got tired of dealing with the weird mechanics cookie-based logins have to deal with, where standard_html_error and tracebacks intrude on login intercepts. I came up with the following minor change, which works great for me, and might make GenericUserFolder's job easier, too. In ZPublisher/HTTPRequest.py (or SiteAccess/ChangeBehaviors.py, if you use SiteAccess), change the lines: if user is None and roles != UNSPECIFIED_ROLES: response.unauthorized() to: if user is None and roles != UNSPECIFIED_ROLES: object = response.unauthorized Under normal circumstances, this has no effect, since 'response.unauthorized' will be called as soon as it is returned and will raise its exception as usual. Now however, we add a twist; When the cookie-based authenticator fails, it replaces 'response.unauthorized' with its 'login' document. If no higher-level authentication succeeds, the 'login' document is rendered normally. Since 'response' is re-created with each request, this modification has no effect beyond the current request. What do you folks think? Cheers, Evan @ 4-am

1 0

Factory pollution, scopes, zope hosting
by Lalo Martins 12 Jan '00

12 Jan '00

I already described a problem of ``factory pollution'' I'm facing. In short, I develop a ZClass for a very specific problem, and then have to cope with its metatype in the ``add object'' lists of the entire site. This is a pain for hosting providers for two reasons: 1: it pollutes the ``add object'' list. I will soon have a half hundred metatypes. It's bad to make customers understand this. 2: perhaps I developed some class for one customer alone. For the others it's a waste of mental space. Perhaps I even want to charge for specific services (like, say, a shopping cart class). I was going to settle for deleting the Factories and making some scripts to add all instances programatically, but then I realized there is a possible sollution. I'd like to hear comments on it from folks with a say on Zope internals before I move on. This sollution is giving a ``scope'' property to the Factory object. This property is a list of strings (represented as ``lines'') defaulting to ['/']. A given factory is only valid in places where the path start with one of the scope strings (so the default scope says this factory is valid anywhere, but I could change it to ['/BroDar/nucleos', '/lalo/foobar'] so it would stop all other places from having this metatype on their ``add object'' lists). Is it worth? Is it a good sollution? Is the problem real, or in my mind? []s, |alo +---- -- I am Lalo of deB-org. You will be freed. Resistance is futile. http://www.webcom.com/lalo mailto:lalo@webcom.com pgp key in the web page Debian GNU/Linux --- http://www.debian.org Brazil of Darkness -- http://zope.gf.com.br/BroDar

1 0

Re: [Zope-dev] simple revision control(?) for Data.fs.
by Anthony Baxter 12 Jan '00

12 Jan '00

>>> Evan Gibson wrote > What if someone packed their database and then did heaps of changes so that > the size was bigger than it was before the packing? > You'd have to add extra checking that, if it was bigger, the first "size" > bytes are the same in both files. ooo. nasty. hm, ok, so I need to check for the datestamp on Data.fs.old Or maybe md5sum the bytes... hmmmm. Anthony

1 0

RE: [Zope-dev] RE: More on the Absent id bug
by Michel Pelletier 11 Jan '00

11 Jan '00

> -----Original Message----- > From: Jason Spisak [mailto:jason@mtear.com] > > I commented out that line, and it still has the backslash in > front of the > catalogued url. > > Michel, is that right? > > Aren't they supposed to not have the leading slash? If it > has a leading > slash, then when you click on the link in the Catalog > management interface, > you get and error. (it is looking for a server @ > http://my_object/url/object, rather that > http://localhost:8080/my_object/url/object. > > Which one is correct in in your design? The latter, obviously. Not putting the slash in confuses your browser into thinking it's a bad relative URL. The culprit is in lib/python/Products/catalogView.dtml Here is the checkin message: ---------------------------- revision 1.10 date: 1999/12/28 23:59:10; author: amos; state: Exp; lines: +1 -1 Fixed links to catalogged objects. ---------------------------- Here is the diff: *** /tmp/catalogView.dtml.~1.9~1NAywY Tue Jan 11 11:24:02 2000 --- /tmp/catalogView.dtml1NA_6e Tue Jan 11 11:24:02 2000 *************** *** 48,54 **** </TD> <td valign="top"><dtml-var meta_type></td> <td valign="top" align="left"> ! <a href="/<dtml-var "getpath(data_record_id_)">/manage_workspace"><dtml-var "getpath(data_record_id_)"> <dtml-if title> (<dtml-var title>)</dtml-if></a> </td> </tr> --- 48,54 ---- </TD> <td valign="top"><dtml-var meta_type></td> <td valign="top" align="left"> ! <a href="<dtml-var "getpath(data_record_id_)">/manage_workspace"><dtml-var "getpath(data_record_id_)"> <dtml-if title> (<dtml-var title>)</dtml-if></a> </td> </tr> As you can see what probably happened was that not slash before the <dtml-var getpath...> probably broke the catalog's ability to show good links in certain circumstances, which fixed one problem but manifested yours. The quick fix is to remove the / in the <a> tag in this file. But we MUST think more about this path stuff, Catalog was the first app to use it extensivly and our lack of a path framework shows. -Michel

3 2

RE: Uncatalog absent id bug - one possible cause
by Michel Pelletier 11 Jan '00

11 Jan '00

> -----Original Message----- > From: Itamar Shtull-Trauring [mailto:itamars@ibm.net] > Sent: Tuesday, January 11, 2000 7:18 AM > To: zope-dev(a)zope.org > Cc: michel(a)digicool.com > Subject: Uncatalog absent id bug - one possible cause > > > Good morning Michael, > > I'm also encountering the "Uncatalog absent id" in a bad way > (users can't > enter data cause reindex fails), and I found at least one > reason why, in > addition to the slash problem (which I deal with by trying to > uncatalog both > with and without a slash.) > > I have a folder with a space in it - "Do Ahead". When I > index using the > "Find objects to index", they get indexed as "/Do > Ahead/object". When I > try to unindex them it tries to unindex "/Do%20Ahead/object", > and then that > error comes up. I guess here is a good example of where it makes sense to raise an error, otherwise it would have silently *not* unindexed your object. > > I added it to the Collector. Thanks, I'm guessing it's a simple fix of adding url_quote somewhere. -Michel > > -- > Itamar S.T. itamars(a)ibm.net >

1 0

RE: [Zope-dev] RE: More on the Absent id bug
by Michel Pelletier 11 Jan '00

11 Jan '00

I'm against doing this off hand without thinking about the problem, according to Jason this never happened before a certain recent time, but neither this code nor any code in Catalog changed at that time. This sort of makes me think this is not the problem. How is it, exactly, that this problem manifests itself for both of you? How are your objects getting cataloged? What changed when the problem started apearing? -Michel > -----Original Message----- > From: David Kankiewicz [mailto:kankie@thegrid.net] > Sent: Monday, January 10, 2000 7:09 PM > To: Michel Pelletier > Cc: 'Jason Spisak'; zope-dev(a)zope.org > Subject: Re: [Zope-dev] RE: More on the Absent id bug > > > Michel Pelletier wrote: > > > <cut> > > > > > Can you open a collector ticket? I'm exploring catalog issues right > > now... > > > > -Michel > > > > OK OK... About the leading slash ////// > > In ZCatalogAware.py > in function url() > > We have: > ... > if uri[0]=='/': uri=uri[1:] > return uri > > > If I comment out the "if", the ZCatalogAware subclassed > Zclasses catalog > themselves correctly, like Find's "/..." :) > > Now, what I want to know is, will this somehow mess up something else? > (I can't see how it would) If not PLEASE check it in to the cvs! > > Bye for now, > David >

1 0

RE: [Zope-dev] ZFormulator review...
by Martijn Pieters 11 Jan '00

11 Jan '00

> -----Original Message----- > From: Chris McDonough [mailto:chrism@digicool.com] > > I'm working on a product that does a similar thing (except > the object is > a "font manager"). I'd be very interested to see how you're doing it > Martijn... > Here's what I do: def initialize(context): # Add the scheduler to the Zope root, and make it undeletable. # I know, this is an extreme hack. But supplying a patch for # OFS/Application.py is just not practical. # We also add a hook to the Product object, so we can make # the scheduler object removable again when this Product is deleted. app = context._ProductContext__app # Don't you love private name mangling if not hasattr(app, 'Scheduler'): oScheduler = ZScheduler.ZScheduler() app._setObject('Scheduler', oScheduler) prod = context._ProductContext__prod setattr(prod, 'manage_beforeDelete', beforeDeleteHook()) get_transaction().note('Added Scheduler') get_transaction().commit() if 'Scheduler' not in app._reserved_names: app._reserved_names = app._reserved_names + ('Scheduler',) # initialize ZEvent objects context.registerClass( ZEvent.OneTimeZEvent, permission = 'Add ZEvent', constructors = ( ZEvent.manage_addOneTimeZEventForm, ZEvent.manage_addOneTimeZEvent), icon = 'www/OneTimeZEvent.gif') class beforeDeleteHook(Persistent): """ Make sure that the Scheduler object can be deleted This should be persistent.... """ def __call__(self, prod, products): app = products.aq_parent.aq_parent rn = list(app._reserved_names) try: rn.remove('Scheduler') app._reserved_names = tuple(rn) except ValueError: pass The beforeDeleteHook object makes sure that if you _manually_ delete the ZScheduler Product in the management interface, the protection on the Scheduler object is removed. As this code is removed when you delete the actual Product subdir from the filesystem this does not get executed when you uninstall your Product that way. -- 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

CORBA Integration Redux
by Tres Seaver 11 Jan '00

11 Jan '00

I have added a couple of documents (or a least starts at them) to the requirements document I posted about last week -- here is the whole set: * "Requirements":http://www.zope.org/Members/tseaver/CosNaming/Requirements * "Roadmap":http://www.zope.org/Members/tseaver/CosNaming/Roadmap * "CORBA @ 30,000 feet":http://www.zope.org/Members/tseaver/CosNaming/BirdsEyeCORBA Feedback welcome! Tres. -- ========================================================= Tres Seaver tseaver(a)palladion.com 713-523-6582 Palladion Software http://www.palladion.com

1 0

RE: [Zope-dev] How do index the contents of a File?
by Michel Pelletier 11 Jan '00

11 Jan '00

> -----Original Message----- > From: Evan Simpson [mailto:evan@tokenexchange.com] > Sent: Friday, January 07, 2000 1:50 PM > To: bauer(a)atlas.unisinos.br; zope-dev(a)zope.org > Subject: Re: [Zope-dev] How do index the contents of a File? > > > From: Carlos Henrique Bauer <bauer(a)atlas.unisinos.br> > > I created a ZCatalog to index all the contents of a site. > To index the > contents > > of the documents I added a 'raw' text index to the catalog, > but it seems > to be > > working just with DTML Documents. Do I need to add another > index to the > catalog > > to make it index File documents or is it a ZCatalog bug? > > File objects are not 'documents', although they may of course > contain files > which you consider to be documents. Since File objects can contain > *anything*, including binary executables or images, they don't define > 'contents' for ZCatalog purposes. Well stated Evan. File's do have an attribute called 'data' that ZCatalog will use like any other attribute (if you have an index called 'data'), however, as you pointed out, you may end up indexing quite a bit of garbage. > What you might want to do is create a sub-ZClass of File dedicated to > storing your document files, and define a > PrincipiaSearchSource method which > returns the contents of the object. You could even > pre-process the contents > in this method if you wanted to strip off headers or > formatting or any other > bits you don't really want in your full-text search. This is probably the best solution. -Michel

2 1

Uncatalog absent id bug - one possible cause
by Itamar Shtull-Trauring 11 Jan '00

11 Jan '00

Good morning Michael, I'm also encountering the "Uncatalog absent id" in a bad way (users can't enter data cause reindex fails), and I found at least one reason why, in addition to the slash problem (which I deal with by trying to uncatalog both with and without a slash.) I have a folder with a space in it - "Do Ahead". When I index using the "Find objects to index", they get indexed as "/Do Ahead/object". When I try to unindex them it tries to unindex "/Do%20Ahead/object", and then that error comes up. I added it to the Collector. -- Itamar S.T. itamars(a)ibm.net

1 0