- Zope-Dev - Zope lists

Reminder: bug day tomorrow Oct 21
by Brian Lloyd 20 Oct '03

20 Oct '03

Hi all - A quick reminder that we'll be having a bug day tomorrow 10/21 to try to knock out some remaining issues in the drive to get a Zope 2.7 final out. Hope to see you on #zope-bugday ! Brian Lloyd brian(a)zope.com V.P. Engineering 540.361.1716 Zope Corporation http://www.zope.com

3 3

__before_publishing_traverse__ problem (2 args only)
by Tobias Herp 20 Oct '03

20 Oct '03

Hi, I have a Python product which defines a __before_publishing_traverse_ method. At last I thought it would all work fine now... Now I added one of my User Folder objects and got an exception because __before_publishing_traverse__ was called with only two arguments instead of three. Quite curious (because of course I don't call this method myself) I had a look at ZPublisher/BaseRequest.py, line 251, and noticed it is called there without providing the REQUEST. When I monkeypatched this method to hand over the request as the third argument, Zope complained that the "global name 'UNSPECIFIED_ROLES'" was not defined (and nothing worked anywhere anymore). So should I make the REQUEST an _optional_ argument to my __before_publishing_traverse__ method? But the purpose of my b.p.t. method is to remove up to two path elements from REQUEST['TraversalRequestNameStack'] and set request variables appropriately, so what is the purpose of this call in BaseRequest? Is there an article anywhere which explains the whole topic in an understandable way? Thanks, Tobias (Zope 2.6.2,

2 1

Making Zope accept international characters in Id's
by Lennart Regebro 20 Oct '03

20 Oct '03

Hiya everyone! I didn't get any response on my last post, so I'm charging ahead full speed. The fact that litmus complains about Zope's inability to accept UTF-8 filenames is something I view as a bug. It can be fixed simply by changing the default _checkid() implementation to allow extended characters. However, I don't want to break anything, so I think it should be optional, controlled by a setting. Either the already existing, but seemingly undocumented setting management_page_charset could be used to check that the given Id is acceptable in that charset. OR, a separate setting could be used, maybe allowed_id_charset? I'm gonna go with that last one if nobody complains. Suggested code (only the important parts): # Any other characters that should NEVER be allowed?bad_id=re.compile(r'[:;/\\]').search def checkValidId(self, id, allow_dup=0): # So should be allow unicode, then? if not id or not isinstance(id, StringType): if isinstance(id, UnicodeType): id = escape(id) raise BadRequestException, ('Empty or invalid id specified', id) # Here is the characterset checking. This works fine, I've tried it. try: charset = getattr(self, 'allowed_id_charset', 'ascii') id.decode(charset) except: raise BadRequestException, ( 'The id "%s" is not of the allowed character set (%s).' % \ (escape(id), self.allowed_id_charset)) # And here we check for unallowed special characters. if bad_id(id) is not None: raise BadRequestException, ( 'The id "%s" contains characters illegal in URLs.' % \ escape(id)) # And then comes the rest of the tests, as usual, for aq_ and such. # I won't copy them in in this mail. Opinions?

3 3

Problem with request/HTTP_X_FORWARDED_FOR
by Fernando Martins 19 Oct '03

19 Oct '03

Hello, Sorry for the cross-posting but I've posted before on the zope list without reply. The following looks like to be a bug on the lookup of the request attributes. If request doesn't have the attribute HTTP_X_FORWARDED_FOR, the following TALES expression should return REMOTE_ADDR. In fact, it returns the empty string. <p tal:content="request/HTTP_X_FORWARDED_FOR | request/REMOTE_ADDR"></p> Naturally, the following works ok: <p tal:content="python: request['HTTP_X_FORWARDED_FOR'] or request['REMOTE_ADDR']"></p> Shall I file it as a bug or am I missing something? Cheers, Fernando

1 0

SESSION loosing data when system has multiple users
by Steve Jibson 17 Oct '03

17 Oct '03

We have been using Zope 2.7.0-beta2 and we have been having a bit of trouble with SESSION loosing data when we have more than one user on the system. These problems usually show up as KeyErrors when a particular data element in SESSION cannot be found. If I set the zserver threads to 1 in zope.conf, the problem seems to go away (but the system isn't really usable either). So I assumed it was a thread-safety problem in the SESSION code. I decided to be brave and take a look at the code. To my surprize, I found a test file called stresstestMultiThread.py in the lib/python/Products/Sessions/stresstests directory that seemed to be the exact thing I wanted to test. It wouldn't run without a couple of minor modifications, but once I got it running, with more than three or four threads, the tests fail (never complete). Is this test file still valid? It looks like it hasn't been modified for almost a year now. If it is valid, has anyone looked into getting the tests to pass. I will probably look into it more myself, but before I do, I just want to make sure I'm not headed down a dead-end street. Thanks, Steve P.S. - The tests failed in both the 2.7.0b2-src and the what is currently in cvs.

2 3

Post-Traversal Hook (+Post-Publication Hook)
by Christian Theune 17 Oct '03

17 Oct '03

Hi, I'm cleaning up the alternative BaseRequest for Zope 2.7 right now which implements a post traversal hook (which will go into 2.8) and discovered a small issue I would like to get some feedback from. Originally the patch called all registered methods after the user has been determined, but before eventually unauthorized is thrown. Therefore the SecurityManager already could deliver the user object, but it is not available within the REQUEST as the well known "AUTHENTICATED_USER". Anyway, does anybody see a special reason to call the hooks a) before unauthorized b) after complete successful authorization I see that there could be two hooks make sense in there, but can't decide for a single one. (My tummy tells me to go for the one after complete authentication, so we have the complete REQUEST environment as in opposition to site access rules.) Also Dario Lopez-Kaesten came up with a request for a hook that is called after the publication. Is there anybody objecting such a "post publication hook"? Otherwise I would go for - implementing post-traversal variant b - implementing a post-publication hook Cheers, Christian PS: Looks a bit like I should write a proposal on this ...? -- Christian Theune, gocept gmbh & co. kg http://www.gocept.com - ct(a)gocept.com fon: 03496 3099112 fax: 03496 3099118 mobile: 0179 7808366

3 4

Zope Financials
by alan 16 Oct '03

16 Oct '03

I know that you are all focused upon content management systems here at Zope/Plone - and with good reason. However, our community has a dearth of commercial products and Java offerings like Compere ERP, and JBOSS are stealing a march in this arena. I've made an initial attempt to redress this by applying the Zope architecture to a general ledger product, BastionLedger, and to an embryonic financials gateway, BastionBanking. I've seen a couple of other Zope products dealing with banking, and would like to somehow coordinate our efforts. This is a very serious initiative, and I've put almost a year's personal work and finance into this product and associated infrastructure. Now, I'd like to see some serious collaboration to get a fantastically fully featured ledger available for Zope clients. What I've delivered so far is quite good in terms of basic functionality and a rich toolkit from which to develop accountancy processes - but payroll systems for example have an abundance of nuances and exceptions and needs much more resource than I have available. In order to overwhelm the world with this stuff, I think we need to start very small. If YOU are a contracter - have your own business, or you've got a Plone site for a club or society, how about getting on board and helping to get the ledger to meet your financial requirements. The general ledger and order management/inventory control already support large-scale use. These products are available for comment and download at http://www.last-bastion.net/BastionLedger and http://www.last-bastion.net/BastionBanking. There is much more information there about exactly what they do and what my current thoughts on this initiative are. Please feel free to discuss them with me on or off this list, or on their own lists. I look forward to some interest. Alan

1 0

Zope, WebDAV, UTF-8 and litmus
by Lennart Regebro 16 Oct '03

16 Oct '03

I've been fiddling around with Zope and WebDAV and have some questions. 1. UTF-8 support. WebDAV requires UTF-8 support. I've been using "Litmus" (http://www.webdav.org/neon/litmus/) to test the Zope access, and by default it doesn't accept UTF-8 names. How does people for example in China and Japan deal with this? I have bypassed this by making a special version of _checkId() that accepts them, but maybe there is another way? If not, I think there should be a way. 2. Other Litmus testing. Of course there are several other failures, besides the UTF-8 support. Has anybody else looked into this, and done anything on it? I could probably spend some bug day time on it, but I don't want to run into the same walls as others already have, if I can avoid it. :)

1 0

WebDAV and UTF-8 filenames
by Lennart Regebro 15 Oct '03

15 Oct '03

Hi guys! I've been hacking around a bit with Zope to try to support some non-ascii characters in the id's. Yes, I know it looks ugly in the URL's but when you go via WebDAV, that is not an issue. :) Anyhow, getting your Zope-object to accept non-ascii characters is not that hard, of course. But the problem comes when you try to use OpenOffice, and do a "Save As...". It seems like OpenOffice sends the new name as UTF-8, translating a nice "filé.doc" to "filÃ©.doc". Of course, OpenOffice will translate it back, so it looks good in OpenOffice, but that the id is different from the webclient and OpenOffice is not a good idea. Could somebody help me shed some light on some issues here: 1. Does this happen because Zope missed out on UTF-8 decoding and encoding somewhere? It seems to me that this should be done in the ZPublisher somewhere, but it isn't, right? 2. I have tried understanding the ZPublisher before, but failed. Can somebody point me to where this decoding encoding should be done, maybe? Or are you of the opinion that it's to complicated a bug to fix? 3. Will Zope 2.7 be Python 2.3 only? Because that would probably make UTF-8 support easier. //Lennart

3 3

Very fast requests beating ZODB commits...
by Richard Jones 15 Oct '03

15 Oct '03

I'm seeing the following behaviour in Zope2.7b2 (python 2.3.2) on a relatively fast computer (Athlon 1.8GHz): 1. request comes in which modifies ZODB 2. code handling request replies with REQUEST.RESPONSE.redirect() 3. redirected page uses data input at step 1 (specifically, it's auth info) but that info hasn't been committed yet, so we get errors So I've modified the method at step #1 to get_transaction().commit() before doing the redirect(), but I'm usually a little concerned when I have to invoke the transaction machinery directly like this... Any thoughts? Is there some way to hold the redirect off until after the transaction is committed normally? Richard

3 3