February 2003 - Zope

zope using up 90+% of CPU
by george donnelly 17 Feb '03

17 Feb '03

hi I have a Zope instance that 3 times in the last 24 hours has gotten to a point where its using up almost all of the CPU. Can someone point me to a good tool or method for figuring out what's causing this? I've got CallProfiler installed now and am building up some data, waiting for this to happen again so I can see if I can pinpoint it but I'm not certain that this is right/best tool for the job. versions... Zope 2.6.1 with Python 2.1.3 (with thread stack size upped to 1MB) on FreeBSD 4.7 and many products installed thanks! <--> george donnelly - http://zettai.net/ - "We Love Newbies" :) Zope Hosting - Dynamic Website Design - Search Engine Promotion Yahoo, AIM: zettainet - ICQ: 51907738 - e:george@zettai.net

3 3

Returning an image and text from python script
by Sven Hohage 17 Feb '03

17 Feb '03

I'd like to return from python script an image object to a dtml method which renders the image. If the state is edit - set by a cookie- the image should have a link for updating the image. I tried to add the text with the object which obviously failed. I also tried to return a string like "<dtml-var '_.getitem(' " + pictureName + " ,0'>" + linkText which only displays the string but didn't render the image. To return multiple values is not what I want because I want to have a simple funtion in the dtml-method like <dtml-var "getPicture('pic1')"> Thanks

3 2

How to use cookieless sessions with using classic HTML URL links?
by Lubos Culen 17 Feb '03

17 Feb '03

Hello, another (maybe newbie?) question: I have developed pages with lot of absolute and relative links, which look like classic HTML links (e.g. <a href="/path/to/file">File</a> or <a href="file">File</a> etc.). Now I added the sessions using (user login/registration) to these pages and want to hold the sessions also for browsers which have disabled cookies. I noticed browser_id_manager is able to do this, but I was able to do it only using absolute_url method written inside each URL - which would mean that I would have to change all the links in all my HTML scripts and modify all the link code, to be able to transfer session data! So the question is - is it possible to maintain sessions in non-cookie way (through URL parameters), while the links in sources would have classic URL format (<a href="file"> or <a href="/path/to/file"> etcc.) and Zope would encode the URLs automatically? Thanks, Lubos.

2 1

Zope and security
by Sorin Marti 17 Feb '03

17 Feb '03

Hi all, I am very new to zope and I had a look at it the last days. In my company we want to use zope for the intranet (for sure) and maybe for our official website. Now I want to know more about the security in zope. How secure is the whole stuff with the users? If employees should be able to see stuff others shouldn't see how secur is it then? any suggestions? any links? any comments? Thankx in advance Sorin Marti

2 1

Adding Role Programmatically
by Samir Mishra 17 Feb '03

17 Feb '03

Hello all, This is probably simple, but I'm unable to find anything on it. I'm trying to add a new role to a new folder using a Python script, and am using - context.manage_addFolder(fldr, folders[fldr], 1, 1) # folders is a dictionary # create a new folder named 'fldr' newFolder = getattr(context, fldr) # create reviewer role in folder newFolder.manage_addRole('Reviewer') The last line generates the following error - Line 20: "_addRole" is an invalid attribute name because it starts with "_". I've looked at the source code in Roles.py but can't find an alternative. What should I be doing here? Is there any documentation on this (I can't find anything in the Zope Book either). TIA Samir

2 1

ZEO client updates from ZODB after sync?
by Paul Winkler 17 Feb '03

17 Feb '03

we use zsyncer to push from a dev machine to a production machine. We sometimes see this behavior: 1) zsyncer reports success. 2) going to the production machine and viewing the management edit form for the object in question shows the old version. 3) eventually the new stuff seems to come up. The interesting thing is that the production system is simple ZEO cluster. All the ZEO clients are exposed to the web via a loadbalancer which presents a single service address to the world. Zsyncer is pointed at the service address to increase the chances that there's always a live server there. But this means that we don't know which production ZEO client actually receives the sync request. I suspect that what happens is: * we sync to http://production/our_zsyncer * we actually sync via the loadbalancer to http://production_1/our_zsyncer * we go to view it at http://production_2/our_zsyncer * we see the old data for a while So I'm wondering: how does a ZEO client know when to reload the object from teh storage? Are there situations in which this doesn't happen "right away"? -- Paul Winkler http://www.slinkp.com Look! Up in the sky! It's TENSILE CONJURER STOOL! (random hero from isometric.spaceninja.com)

2 3

Problem to log out from CMF Site
by Mico Siahaan 17 Feb '03

17 Feb '03

This is another confusing thing for me. I tried to installed CMF and add CMF Site. Everything seemed fine until I tried to log out from the site. I got error message: 'Site Error. An error was encountered while publishing this resource. Error Type: Bad Request error Value: Invalid authentication token' Because I am a zope newbie, I wonder what caused this. How to solve it? I'm using Zope 2.6.0 binary release, linux Mandrake 8.2 , CMF 1.3. Thanks. -- mico(a)cbn.net.id --- Ph. 62-22-2505549 ICQ #12521756

2 1

RE: [Zope] Re: [Zope-mosx] Local FS files "not found"
by baiewola＠netscape.net 17 Feb '03

17 Feb '03

Here's an example of my problem: http://spider.engin.umich.edu:8080/testing/index_html This is a test LocalFS that I set up just now. The headers probably won't show up for you, but as you can see, the documents in my file system are actually listed on the page, they simply can't be downloaded. Here's something I just noticed: Instead of using the basepath I defined under "Properties" in the Local FS object called "testing" to display the URLs of the items on my hard drive, Zope is using a path relative to /testing/ To put it more succinctly, Zope appears to be ignoring the basepath property. Steve Spicklemire <steve(a)spvi.com> wrote: >What's the URL she actually clicks on? Does it look reasonable? > >Have you tried tcpwatch? http://hathaway.freezope.org/Software/TCPWatch > >-steve > >On Monday, February 17, 2003, at 11:19 AM, baiewola(a)netscape.net wrote: > >> I've had Local FS installed for a year now and finally when someone >> tried to use it, it stopped working! >> >> The hard drive location of the LocalFS is >> /Library/WebServer/Documents/SomeFolder >> I've given ug+rwx permissions to SomeFolder. >> >> In Zope, the basepath for the LocalFS is defined as >> /Library/WebServer/Documents/SomeFolder >> >> When the user goes to the URL of this LocalFS, she sees a list of PDF >> files in the file system, but when she clicks on one (or >> control-clicks to download) she gets a "not found" error: >> >>  >> >> However, the documents she is looking for DO exist in the correct >> location on the hard drive. They also have ug+rwx permissions, and she >> is a member of the appropriate group. >> >> I've tried my programmer's suggestion of adding a / after SomeFolder >> in the basepath property, but that did not work. >> >> Any tips how to start sorting this out? I'm using Zope 2.5.1 on Mac OS >> X 10.2.3 >> >> __________________________________________________________________ >> The NEW Netscape 7.0 browser is now available. Upgrade now! >> http://channels.netscape.com/ns/browsers/download.jsp >> >> Get your own FREE, personal Netscape Mail account today at >> http://webmail.netscape.com/ >> _______________________________________________ >> Zope-mosx mailing list >> Zope-mosx(a)zopeonarope.com >> http://www.zopeonarope.com/mailman/listinfo/zope-mosx > > >_______________________________________________ >Zope maillist - Zope(a)zope.org >http://mail.zope.org/mailman/listinfo/zope >** No cross posts or HTML encoding! ** >(Related lists - > http://mail.zope.org/mailman/listinfo/zope-announce > http://mail.zope.org/mailman/listinfo/zope-dev ) > __________________________________________________________________ The NEW Netscape 7.0 browser is now available. Upgrade now! http://channels.netscape.com/ns/browsers/download.jsp Get your own FREE, personal Netscape Mail account today at http://webmail.netscape.com/

1 0

MVC and Zope
by Kevin Carlson 17 Feb '03

17 Feb '03

Hi everyone, I posted an email yesterday about a problem I had forwarding a request from a python script to a dtml-document. The problem turned out to be centered on the way that the relative URLs were generated. This got me thinking about a larger issue, that of implementing a model-view-controller type of architecture within Zope. Lengthy email to follow, bear with me... If you use a python script as your controller and read necessary configuration information from the traverse_subpath binding, you could set up a table of target dtml documents (or methods), add the necessary model elements to the request and pass the entire thing off the the view for processing. This works great in java, but because of the way that acquisition is handled in scripts, it seems to be an issue in Zope, although I am certain there is a way to code around the limitations. The problem with doing this is primarily with images, stylesheets, and anchor tags that are generated from within the view (or dtml-document/method). If I access a URL like: "www.domain.com/search?arg=Whatever", the system clearly navigates to the search page and can get the arg from the request. If I format the URL as "www.domain.com/controller/search/Whatever" the system will give control to the controller script, which can then see that the search page needs to be called with the argument of 'Whatever'. The problem in doing this is that the relative URLs generated by the search page will then include a path to the controller script. Of course something like, <img src='images/go.gif'> gets returned to my browser with an absolute path of www.domain.com/controller/images/go.gif, which does not get found (the image is actually at www.domain.com/images/go.gif) I believe this is primarily due to the fact that scripts "intercept" control and pass the less significant path elements onto the traverse_subpath binding, which of course, stops acquisition in its tracks. Whew! After that long explanation I can finally get to the question... Is there a way to read the traverse_subpath and if it contains certain text, change the context of the call and return a different object? I noticed a few things in the archives regarding MVC, but didn't notice any concrete solution (yet). Has anyone experienced this before or tried to implement MVC within Zope? Thanks, Kevin

3 3

acl_users and cookie crumbler login problem
by philrobinson 17 Feb '03

17 Feb '03

Here's my ZODB structure: parent_folder -- child_folder_1 -- child_folder_2 I want users to be able to authenticate into each of the child folders independently, so I put an acl_users and cookie crumbler into each of the child folders (can be either standard user folder or XUF, makes no difference). I then define a single user in each child folder as follows: 'child_1' in 'child_folder_1/acl_users' 'child_2' in 'child_folder_2/acl_users' I set each cookie crumbler to 'Use cookie paths to limit scope'. I then put a file called 'login_stub' in each of the folders with its 'view' acquisition turned off. Point a browser at either of the 'login_stub's, and you get a login form; enter a valid name/pwd for that folder and you have access (of course, you had view access before, but now you have whatever rights the relevant user has, which may be more than just 'Anonymous'). Looking at a different child folder does the same thing again as you would expect. Inspecting the cookies shows their paths to be '/child_folder_1' etc. So, this gets me the ability to authenticate into each child folder separately, perhaps using a different user name. If I put an acl_users and cookie_crumbler into 'parent_folder' (to allow a different group of people to manage at that level), however, it all goes wrong. With these present, the cookie paths from the above example become '/', which seems to indicate that they are being generated in the parent_folder rather than the relevant child one. And login fails; presumably because the username supplied for (say) 'child_folder_1' is not valid for 'parent_folder'. If I remove the acl_users and cookie_crumbler from parent_folder, the child folders work fine again. To summarise: if the higher level ones are present, zope uses them. If not, it uses the child one as it should. Why should the presence or absence of the objects in 'parent_folder' affect the child's behaviour? I don't see how it can be caused by restrictive permissions on something in the child folders, since it wouldn't work without the parent_folder objects, but this seems to be kind of the opposite of what acquisition is supposed to do. Anyone run into this kind of thing or am I just trying to be too fine grained and it can't be done? Thanks in advance.... Phil Robinson PS Without the cookie crumblers present, each child seems to behave as you expect. Of course, you can only be logged into one at a time, but the presence or absence of an acl_users in 'parent_folder' doesn't affect their behaviour.

2 1