March 2005 - Zope-Dev

Zope on Windows - REQUEST for volunteers
by Andreas Jung 21 Mar '05

21 Mar '05

Hi, there are several open issues in the Zope collector with Zope running on Windows. I can't see anyone worrying about the Windows platform (I also don't care about Windows as UN*X guy). So if you are interested in running Zope on Windows in a reasonable way than *you* should help to fix the outstanding bugs. I heard some remarks (behind my back) that Zope is not a stable solution on Windows...possibly.....but since Zope is a community-driven project people doing such remarks should either shut up or help to improve the situation by volunteering (which is also true for other parts of the Zope development). So if you want to improve running Zope on Windows: go the Zope collector, search for Windows related problems, find a solution, submit a patch and don't expect that everything can be handled by the handful of active Zope 2 core developers...means it's also up to *YOU*. Andreas

6 6

[Zope 2.8] release schedule
by Andreas Jung 20 Mar '05

20 Mar '05

I have updated the Zope 2.8 wiki to reflect the planned schedule for Zope 2.8: http://www.zope.org/Wikis/DevSite/Projects/Zope2.8/MilestonePlan Since we are trying to shift towards a fixed-date schedule every Zope 2 contributor should see this schedule as a chance and commitment to make contributions and fixes within this time frame. This schedule implies that all outstanding issues with Zope 2.8 have to be fixed before the final release. Andreas

1 0

Cookie problem with mySqlUserFolder
by Aruna Kathiria 19 Mar '05

19 Mar '05

Hello everyone: I am facing the problem regarding cookie. One of our website uses mySqlUserFolder. Problem is if user logs on one computer, and if he/she comes again on same computer, computer remembers him/her. That is good as it is what we want. But if he/she goes on another computer and logs on and then if he/she come back to previous one, it looses it's reorganization though cookie is there. I checked what problem would be and found that each time logged in process on different computers generates a random value for cookie on that computer and also store same value in Tokens table on mysqlUserFolder database. So everytime only cookie from one computer matches the value with Tokens tables value. So other computer don't remember the previous login. ( Though I don't know this is the reason of my problem) What would be the solution to keep reorganization on all computers on which once user has been logged in? Any help will be appreciated. Thanks in advance! With Regards, Aruna Kathiriya Sr.Consultant, CIGNEX Technologies, Inc T: 408.327.9900 x 314 F: 408.273.6785 E: aruna(a)cignex.com U: www.cignex.com "Implement IT Right"

2 1

Re: [Zope-Checkins] SVN: Zope/trunk/ - Applied patch for
by faassen＠infrae.com 18 Mar '05

18 Mar '05

On Thu, Mar 17, 2005 at 10:34:10AM -0500, Tim Peters wrote: > [Sidnei da Silva] > .... > > Humm... we are trying to push a Zope 2.8 beta out, do you have or know > > of plans to use ZODB 3.4 with Zope 2.8? > > Yes. Jim needs to fix ZClasses for 2.8 too. ZODB 3.4 requires some > Zope(3) features, like zope.interfaces and zope.testing. Jim is > adding those on his ZClasses branch. When the ZClasses work is merged > (but not before), it becomes possible to move to ZODB 3.4. Zope3 > already uses ZODB trunk (3.4 development). We can't afford to > maintain a separate ZODB just for Zope 2.8, so view moving to ZODB 3.4 > as a requirement for Zope 2.8. Oh, also please note that we're merging zope.interfaces and zope.testing into Zope 2.8 already. So once we merge that branch (today, we expect), ZODB 3.4 could be merged in. But I'm worried that there are other factors involved with ZODB 3.4 that will hold up release even further. For instance, there is no ZODB 3.4 release yet; how much work would that entail? Regards, Martijn

2 1

"Overriding Products"
by Martijn Jacobs 18 Mar '05

18 Mar '05

Hello guys. I have a question regarding product importing in zope 2.7.x, as since zope 2.7 this part is way more flexible then it was before. I'd like to accomplish the next situation : - A Zope base installation has Product X installed in /lib/python/Products which I want to change or develop further - x instances of Zope use the produtc from this base install. - At one instance I want to test and develop, without breaking the other instances. - Copying the Product X folder to the <instance_home>/Products folder is not an option, as import conflict errors occure if you do not remove the product from the base installation. - Copying a local version in all 4 instances is a solution, but to manage that will cost alot of work, especially if you have alot of instances. - What I really want is that you can copy the Product X folder to the Products (or "CustomProducts" for example) of an zope instance, and that zope will figure out itself that it should import that version of the product instead of the copy in the base installation. 'Overriding' base products is the name you could give this situation. I tried alot myself, for example checking in the __init__.py if there is a copy of this product in <instance_home>/CustomProducts and then importing that product instead of my own, but I always get into trouble : AttributeError: 'module' object has no attribute '<Classname>' It doesn't have to work for all products, only for some of the products that we are developing ourselfs. We also have a different structure than normal products as following : Main Product - Sub product X - Sub product Y In the __init__.py of Main product we import and initialize the subproducts. The idea is to put one of those subproducts in a zope instance/CustomProducts folder for further development and that the other instances of zope still uses the base version of the subproduct. I hope you can help, thanks in advance for your time. Kind regards, -- Martijn Jacobs East Site | Postbus 642 | 6800 AP Arnhem t: +31 (0) 26 35 14 295 | f: +31 (0) 26 44 20 432 @: mart(a)eastsite.nl | w: www.eastsite.nl

3 2

Zope 2.8 + Five post-sprint status
by faassen＠infrae.com 18 Mar '05

18 Mar '05

Hey everybody, We're wrapping up here at a very pleasant and productive Zope 2/3/Five sprint here in Paris. We've accomplished quite a lot, and we'll let you hear what this is in more detail soon. We've spent a lot of time with Zope 2.8, integrating Zope X3.0 and Five into it. Our work is on two branches: The five-integration branch: svn://svn.zope.org/repos/main/Zope/branches/five-integration The trunk, which has synchups with Zope 2.7 fixes: svn://svn.zope.org/repos/main/Zope/trunk The Five integration branch contains Five, and a version of Zope X3.0 made for integration with Zope 2.8 (as a svn:external). We hope that this branch can be merged into mainline Zope 2.x trunk soon. I hereby want to thank everybody here at the sprint who worked so hard on this. It's really your work, guys, I just kept the todo list. :) Regards, Martijn

1 0

checkPermission and proxy roles
by Tim McLaughlin 16 Mar '05

16 Mar '05

Issue #'s 78, 846, 1125 all relate to checkPermission not considering proxy roles. #78 was resolved, however, from the looks of the code (and I haven't tested this), it will fail to Cut/Paste because the 'Delete objects' perm would be permitted by the proxy role which is not considered in the checkPermission. Anyway, the question is this... Why doesn't checkPermission consider proxy roles? I found the following discussion in Zope-CMF and it looks like this was "fixed" in CMF, but it seems to me like this should be the default policy behavior in Zope. We can do the patches if there's no reason for it being the way it is... Cheers, Tim ------------------------------------ begin quote Dieter Maurer wrote: > I think, we should have both possibilities: > > [1] check whether the real user would have the permission > (independent of proxy roles) > > [2] check whether the current context has the permission > (dependent on the current proxy roles and other > execution security aspects (such as ownership)) I'd like to replace utils._checkPermission in CMF HEAD with the attached code. This would change the behavior of _checkPermission from [1] to [2]. (If I didn't make a mistake.) The way utils._checkPermission is used in CMF implies possibility [2] would be the right behavior, the fact it implemented [1] looks like a bug to me. I don't know of any code that'll break if we switch to [2]. I can see there might be a need for [1]. But in this case you can use Zope's checkPermission method. If there are no objections I'll soon make a CVS checkin of the attached code. Cheers, Yuppie ----------------------------------- -- Tim McLaughlin Chief Technology Officer Siteworx, Inc. Innovative internet solutions. 703.964.0300 ext. 208

1 0

Re: [Zope3-dev] Re: Heads-up: Zope 2.8, Zope 3 and Five
by faassen＠infrae.com 16 Mar '05

16 Mar '05

On Tue, Mar 15, 2005 at 07:31:51PM +0100, Christian Heimes wrote: > Martijn Faassen wrote: > >[Could you point me to the issue or mail describing the old-style BTree > >problem? I may have run into it under another name or something.] > > http://zope.org/Collectors/Zope/1695 > > Persistent* were fixed yesterday by Tim Peters Thanks for the pointer. [snip] > That sounds a little bit better but Jim was worried for a long time to > add zope.interfaces and related. Just read the last discussion over > adding parts of ZopeX3 to Zope2.8. I was involved with the discussion. It actually turns out stitching in parts of Zope 3 is much much harder than stitching in all of it; the latter is just a distribution issue. > On the other hand we maybe get more developers from Five and ZopeX3 into > Zope2 because they want to use Zope 2.8. > > Let's see what the future will reveal us. Yes. And again, I totally understand where you're coming from. I'm coming from the same direction. Regards, Martijn

1 0

Odd behavior of undoable_transactions()
by Paul Winkler 16 Mar '05

16 Mar '05

Is anyone but me bothered by this behavior of undoable_transactions()? If I have a folder /foo and another folder /foobar, transactions for /foobar are displayed in foo.undoable_transactions(). This is clearly caused by lib/python/ZopeUndo/Prefix.py. The behavior matches the docstring - Prefix only cares if one string starts with the other, nothing more. But is that really desirable? Since this is intended purely for use in Zope, wouldn't it be more useful if it cared about paths as such? Something like (untested): class PathPrefix: """A PathPrefix() is equal to any path string it is a prefix of. Two prefixes can not be compared. """ __no_side_effects__ = 1 # what's this mean anyway? def __init__(self, path): path_as_list = path.split('/') self.length = len(path_as_list) self.path = path_as_list def __cmp__(self, o): other_path = o.split('/') rval = cmp(other[:self.length], self.path) return rval -- Paul Winkler http://www.slinkp.com

2 3

Re: [Zope3-dev] Re: Heads-up: Zope 2.8, Zope 3 and Five
by faassen＠infrae.com 15 Mar '05

15 Mar '05

On Tue, Mar 15, 2005 at 04:57:00PM +0100, Christian Heimes wrote: > I really *love* to have Five and parts of ZopeX3 in Zope2 but I don't > like how it is happening. Zope 2.8 is starting to stablize and still > contains some critical bugs like the incompatibility with old style > BTrees and you are going to add a heavy bunch of new features to it. [Could you point me to the issue or mail describing the old-style BTree problem? I may have run into it under another name or something.] > I would prefere this way: > > * Get Zope 2.8.0 out ASAP > > * Branch of Zope 2.9 from 2.8 when 2.8 is in rc phase. 2.9 should be > the target for ZopeX3 and Five only, no other features so you can get > 2.9 out some month afte 2.8.0. > > "Release often, release early and one step after another". You are going > to make the same ugly and bad mistake we made with Plone 2.0.0. :( I completely understand where you're coming from and I share the worries. Some things mitigate this problem: * integrating Five doesn't actually imply many significant changes to Zope 2.8; it'll just be its own product. * integrating zope 3 only means some integration in the build script for Zope 2. It's just a Python library otherwise. * we're separating the work on Zope 2.8 and Zope 2.8 + Five, so we can back out from this plan if we want to, later, and go the 2.8 and 2.9 route anyway. So, in effect, we're not actually *creating* any new features. We're also not doing any deep integration work (Five already has that work). The features already exist and we're just distributing them along with the standard Zope 2.8 distribution. We're not ambitious about new stuff at all. I hope that is somewhat different from the path Plone took for 2.0. Regards, Martijn

1 0