[Zope] Re: help debugging a "can't pickle" error deep within a catalog reindex
Rob Miller wrote: hi, i'm trying to perform a ZCatalog.refreshCatalog() on a catalog with over 29,000 indexed objects. it churns for a good long time, and eventually fails with a long set of tracebacks, of which i've included a sample at the end of this message. i think i understand the gist of the issue... it's trying to write an object (probably a CatalogBrain) to the database, but this object's __dict__ contains a value that is of type instancemethod, which isn't allowed for persistent objects. the problem is that i can't figure out which specific objects are causing the problem. i've used pdb.post_mortem to get a debug prompt way down in the traceback, but the code goes in and out of C modules, so i'm missing a lot of what's happening. and when i interactively peek at the objects that are being indexed when the error happens, there doesn't seem to be anything wrong, and i can index the objects w/ no problem. i've even tried dropping the subtransaction threshold down to 1, so it will try to commit a savepoint after every object, but none of the objects being indexed seemed to have any problems. okay, after leaving this for a few days, i came back to it and managed to work it out. for posterity's sake, what ended up working was that i added an explicit transaction.commit() after every object reindex, wrapped in a try: except, like so: --- src/Zope/lib/python/Products/ZCatalog/ZCatalog.py 2007-10-29 06:09:30.0 -0700 +++ lib/zope/lib/python/Products/ZCatalog/ZCatalog.py 2008-06-24 10:47:49.0 -0700 @@ -294,6 +294,11 @@ if obj is not None: try: self.catalog_object(obj, p, pghandler=pghandler) +try: +transaction.commit() +except: +import sys, pdb +pdb.post_mortem(sys.exc_info()[2]) except ConflictError: raise except: this made the reindex take a ridiculously long time, and it leaves the catalog in an inconsistent state (i.e. it should only be done on a copy of the database, one that you can throw away), but it did in the end reveal to me which record was causing the problem. the problematic record ended up being a ghosted catalog entry that happened to share the same name as a view. when the catalog tried to look up the object by its path (via unrestrictedTraverse), the view object was returned. thanks for the help! -r ___ Zope maillist - Zope@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 )
[Zope] Re: help debugging a "can't pickle" error deep within a catalog reindex
Ross Patterson wrote: Rob Miller <[EMAIL PROTECTED]> writes: hi, i'm trying to perform a ZCatalog.refreshCatalog() on a catalog with over 29,000 indexed objects. it churns for a good long time, and eventually fails with a long set of tracebacks, of which i've included a sample at the end of this message. i think i understand the gist of the issue... it's trying to write an object (probably a CatalogBrain) to the database, but this object's __dict__ contains a value that is of type instancemethod, which isn't allowed for persistent objects. the problem is that i can't figure out which specific objects are causing the problem. i've used pdb.post_mortem to get a debug prompt way down in the traceback, but the code goes in and out of C modules, so i'm missing a lot of what's happening. and when i interactively peek at the objects that are being indexed when the error happens, there doesn't seem to be anything wrong, and i can index the objects w/ no problem. i've even tried dropping the subtransaction threshold down to 1, so it will try to commit a savepoint after every object, but none of the objects being indexed seemed to have any problems. i CAN verify that the instancemethod that is causing the problem renders like this: > even that hasn't proven enough for me to concretely identify the source of the problem, though. i've been working on this for a full day already, and am not sure how to proceed. does anyone have any debugging tips that might help me figure out what, exactly, is causing the reindex attempts to blow up? thanks! Can you get a pdb.post_mortem prompt at the actual ZODB/serialize.py:416 error or only at transaction/_transaction.py:267 where some sort of previous error is handled? yes, i put the post_mortem around the "transaction.savepoint" call at line 559 of Products/ZCatalog/ZCatalog.py. this is the place where the error is happening, but it's where the subtransaction threshold has been reached, so IIUC the error could be caused by any of the objects in the subtransaction. If the former, I often find it informative to step up to ZODB/serialize.py:407 where obj.__getstate__() is called and find out what obj is. Is it the same object every time? If not, is it of the same type every time? right. as long as the subtxn threshold number is the same, then the object is the same. but if i change the subtxn threshold, then the object will be different. as i said, i even tried reducing the subtxn threshold to '1', so, assuming my understanding is correct, this would be triggered after every object. it took many hours to get to my debug prompt; when i did, the obj was FileAttachment object that didn't seem to have any problems. i was able to interactively index the object from the pdb prompt, and even commit the transaction, w/ no problems. A next step can also be to put a pdb.set_trace() at transaction/_transaction.py:340 in the register() method such that it's is only called when the offending object is added to the transaction. i'll give this a shot, thx. Hope some of that helps, even just having someone to discuss it with is helpful, for my sanity at least... ;-) thx. -r ___ Zope maillist - Zope@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 )
[Zope] help debugging a "can't pickle" error deep within a catalog reindex
hi,
i'm trying to perform a ZCatalog.refreshCatalog() on a catalog with over
29,000 indexed objects. it churns for a good long time, and eventually fails
with a long set of tracebacks, of which i've included a sample at the end of
this message.
i think i understand the gist of the issue... it's trying to write an object
(probably a CatalogBrain) to the database, but this object's __dict__ contains
a value that is of type instancemethod, which isn't allowed for persistent
objects.
the problem is that i can't figure out which specific objects are causing the
problem. i've used pdb.post_mortem to get a debug prompt way down in the
traceback, but the code goes in and out of C modules, so i'm missing a lot of
what's happening. and when i interactively peek at the objects that are being
indexed when the error happens, there doesn't seem to be anything wrong, and i
can index the objects w/ no problem. i've even tried dropping the
subtransaction threshold down to 1, so it will try to commit a savepoint after
every object, but none of the objects being indexed seemed to have any problems.
i CAN verify that the instancemethod that is causing the problem renders like
this:
/session_data_manager>>
even that hasn't proven enough for me to concretely identify the source of the
problem, though.
i've been working on this for a full day already, and am not sure how to
proceed. does anyone have any debugging tips that might help me figure out
what, exactly, is causing the reindex attempts to blow up?
thanks!
-r
Traceback (most recent call last):
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/Products/ZCatalog/ZCatalog.py",
line 296, in refreshCatalog
self.catalog_object(obj, p, pghandler=pghandler)
File
"/home/rob/topp/14000/builds/20080611/opencore/zope/Products/CMFPlone/CatalogTool.py",
line 367, in catalog_object
self._increment_counter()
File
"/home/rob/topp/14000/builds/20080611/opencore/zope/Products/CMFPlone/CatalogTool.py",
line 395, in _increment_counter
self._counter.change(1)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/BTrees/Length.py",
line 55, in change
self.value += delta
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZODB/Connection.py",
line 890, in register
self._register(obj)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZODB/Connection.py",
line 900, in _register
self.transaction_manager.get().join(self)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/transaction/_transaction.py",
line 273, in join
self._prior_operation_failed() # doesn't return
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/transaction/_transaction.py",
line 267, in _prior_operation_failed
raise TransactionFailedError("An operation previously failed, "
TransactionFailedError: An operation previously failed, with traceback:
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZServer/PubCore/ZServerPublisher.py",
line 25, in __init__
response=b)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZPublisher/Publish.py",
line 401, in publish_module
environ, debug, request, response)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZPublisher/Publish.py",
line 202, in publish_module_standard
response = publish(request, module_name, after_list, debug=debug)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZPublisher/Publish.py",
line 119, in publish
request, bind=1)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZPublisher/mapply.py",
line 88, in mapply
if debug is not None: return debug(object,args,context)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/ZPublisher/Publish.py",
line 42, in call_object
result=apply(object,args) # Type s to step into published object.
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/Products/ZCatalog/ZCatalog.py",
line 260, in manage_catalogReindex
self.refreshCatalog(clear=1, pghandler=handler)
File
"/home/rob/topp/14000/builds/20080611/opencore/lib/zope/lib/python/Products/ZCatalog/ZCatalog.py",
line 296, in refreshCatalog
self.catalog_object(obj, p, pghandler=pghandler)
File
"/home/rob/topp/14000/builds/20080611/opencore/zope/Products/CMFPlone/CatalogTool.py",
line 385, in catalog_object
update_metadata, pghandler=pghandler)
File
"/home/rob/topp/14000/builds/20080611/opencore/zope/Products/CacheSetup/patch.py",
line 96, in catalog_object
uid, idxs, update_metadata, pghandler)
File
"/home/rob/topp/14000/builds/20080611/opencore/zope/Products/CacheSetup/patch_utils.py",
line 6, in call
return getattr(self, PATTERN % __name__)(*args, **kw)
File
"/home/rob/topp/14000/builds/20080611/opencore
[Zope] Re: which operating system quandry
David Bear wrote: I know it has been asked many times which os is best for zope, I have yet to see a 'system administration perspective' in the discussion. By this I mean which os seems to have the most worry free administration of a zope instances. I have installed zope on both FreeBSD and various linucies and here is what I have observed: 1) Freebsd ports collection is an easy way to install zope -- the ports maintainer takes care of making the build files to handle all the zope dependencies including python versions, libraries, etc. However, I have yet to use portupgrade to apply security patches to zope instances running on freebsd. Does anyone know of using cvsupdate for getting security patches works over the ports collection of zope smoothly? 2) I've run zope on Red Hat and Suse linux. In both cases I found that I needed to install a different version of python than the one packaged with the distro becuase Zope had specific dependencies for new versions python. Applying patches to zope is manual. It has always been very inconvenient to build python and PIL in a separate run instances for zope. This seems like a major pain in the ... i've been most happy w/ debian or ubuntu, and have also had good luck w/ gentoo. in each case, i use python (as well as nearly all python dependencies, e.g. PIL) from the distribution, but then build Zope itself from source, either a checkout or a tarball. -r ___ Zope maillist - Zope@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 )
[Zope] Re: Using property() function in Zope 2.8
On Thu, 06 Jul 2006 21:19:36 +0200, Max M wrote: > I needed to dynamically generate local roles for an Archetypes based > content object today. > > Different layers in my Plone stack breaks all rules and reads the > __ac_local_roles__ variable directly, instead of calling get_local_roles() > > So to maximize the compatibility between Zopes zmi and Plones local roles > management I wanted to make '__ac_local_roles__' a property with setters > and getters. it's not an answer to your original question (i have nothing to add to what fred already replied) but TeamSpace solves this by using a ComputedAttribute instead of a property for the dynamic local roles. all of the pertinent code is here, hope you find it useful: http://svn.plone.org/view/collective/teamspace/tags/1.4/security.py?rev=24604&view=auto -r ___ Zope maillist - Zope@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 )
[Zope] Re: Groupware in zope
Saura Ramachandran wrote: Hi, Is there a groupware in Zope with features like wiki, forum, filemanager and other project management stuff in zope or plone? Please share your experiences. you might find the Plone-based OpenCore software that is driving the openplans.org site useful. it's feature-lean at the moment (basically just a group-centric wiki), but is under very active development, over the next two months we'll be deploying file attachments, mailing lists, blogs, and project rosters. http://plone.org/products/opencore http://openplans.org/projects/opencore -r ___ Zope maillist - Zope@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 )
[Zope] Re: Presentations Available
Nick Davis wrote: Hello Some thoughts : - Upgrading to Zope 2.8 without reading the release notes. I followed the release notes but this didn't fix catalog errors. I am not the only one. This is apparently fixed in Zope 2.8.2 but that doesn't look like it can be downloaded yet. in some cases the problem will go away if you simply execute a 'len(catalog)' command, either from a script or from zopectl debug. YMMV. - Third party products that are not yet fully compatible with Plone 2.1 and/or Zope 2.8. Should this perhaps be the other way around? If those products used the APIs correctly, perhaps a new release of Zope/Plone should still support old APIs and/or deprecate them gradually and with warning? This is a difficult problem to address but sometimes it may be better to hold back on releasing something new if it causes things that depend on it to break. On the other hand I can understand peoples desire to release something new and cool (even if not quite ready) . ;-) And its hard to know quickly what the bugs are if you don't release it to the real world. So hard to know what to do.. the problem is not with the APIs... those we have taken care to deprecate. the problems lie within template code (such as the problem you describe below), where it's much harder to maintain backward compatibility and still move forward. - Sites that have customized parts of Plone they shouldn't have touched in the first place. This is an important issue. It is difficult not to touch things that one shouldn't. A trivial example - If you want your breadcrumbs to just list the breadcrumbs instead of say "you are here" you have to copy across global_pathbar.pt and take out "you are here". There is not another easy way to do it that I can see. If you then migrate to 2.1, that .pt has changed so you have to copy the new one and re-do it otherwise nothing renders. If you want to add another logo on the right of the header you have to hack Plone's templates further. Each change is in itself trivial but they add up and when you migrate, you;re left with stuff that doesn't work and spend quite a while resolving it. And thats for those of us savvy enough to use the filesystem. Those who customised through the ZMI will have a bigger headache. tres responded to this more eloquently than i'd be able to... as for the broken products, that is a pandemic within the open source community, hardly unique to plone. True. AT's problems are entirely recognized by those of us who use it heavily, and i can assure you that the AT developer pool has no intention of continuing to pile more cruft on top of a shaky stack. . the first iterations of this will probably also have some warts. but please don't assume that plone/AT developers don't see the same problems that you see, and that they aren't willing and able to learn from their mistakes. It would seem Archetypes has improved over time. My real worry is when we do have a new release of Plone sitting on top of Zope 3 we'll have a whole new set of bleeding edge code sitting on top of other bleeding edge code, while stuff that did work with a mature AT1.3.x (or 1.4.x or whatever) suddenly stops working. Hopefully this will prove to be an unjustified fear! as tres said, z3 isn't really bleeding edge any more. i'm not saying that there won't be migration bumps.. i expect there will be. but there are a lot of folks with a lot of working code dependent on this stack, and we'll all be working together to get to the next level. Probably this is no-one's fault. It is the nature of open source. To compare, have to admit I tried to get Bricolage to work a while back, and ran into CPAN dependency hell. I wonder if perhaps the real problem is trying to do so much with so few resources. this is, to me, the crux of the problem. there are a million things that could be better. i've got far more ideas for improvement than i have the time to give those ideas... we've all still got to get billable hours in. that being said, things are getting better with time, and i expect them to continue to do so. Linux has a very mature platform as much of its base, and a lot of commercial support which helps. Perl and the CPAN is quite mature now and also had quite a lot of commercial support. The good thing about commercial support is people being paid to do grunt work and run loads of tests and update documents. Both Linux and CPAN are very modular. It is relatively easy to change one part without knowing about other parts. I think it would be easier to find someone who could patch a broken CPAN module, than someone who could delve inside Zope and Plone. This is because many of us don't understand the various components of the underlying architecture, and a lot is changing for Zope 3. Also many Perl modules are widely used by many applications, whereas a lot of Zope code is only used by other Zope code. again, tres
[Zope] Re: Presentations Available
Chris Withers wrote: Nick Davis wrote: there seem to still be migration problems and broken products which prevent people going to 2.1 yet. Yup. this strikes me as a bit unfair. to quote stefan holek from a post on plone-dev earlier today, most reported migration problems are due to: - Upgrading to Zope 2.8 without reading the release notes. - Third party products that are not yet fully compatible with Plone 2.1 and/or Zope 2.8. - Sites that have customized parts of Plone they shouldn't have touched in the first place. - Plone Team stupidity. thus, while it's true that we have made mistakes (and will continue to, no doubt), most of the problems are issues beyond our control. as for the broken products, that is a pandemic within the open source community, hardly unique to plone. how many zope products are out there of dubious quality, or that are poorly maintained? whenever you get a large community of developers, you get a mixed bag of talent and of follow-through. the suggestion you make in your talk of having a peer rating process for add-on products is a good one, certainly, and one that's been discussed before, but getting there takes some effort. My colleague has spent a long time trying to migrate a Product he wrote, from Archetypes 1.2.5 to 1.3.4, due to the fact he had to hack around problems with references. Archetypes is the chief sinner in all of this, I'm afraid. It's trying to solve a very difficult problem, and one which needs tackling with structure and upfront and intuitive design rather than the organic tacking on of new "bitz" whenever anyone felt like it that AT has suffered through... My fear is as more features are added, what you describe as a shaky stack of complex fragile components will get ever more dependencies and therefore ever more complex and fragile. Yup. yes, AT is in many ways a mess. but, as you say, it's tackling a very difficult problem, and, like most attempts at tackling difficult problems, the first iterations were somewhat less than perfect. the same can be said for zope itself... there's a reason z3 was a complete rewrite. AT's problems are entirely recognized by those of us who use it heavily, and i can assure you that the AT developer pool has no intention of continuing to pile more cruft on top of a shaky stack. instead, we're looking at how we can break the framework apart into components that can all be glued together in a nice z3 fashion. there are a great many tools available to us now that were not available when AT was originally developed (adapters, events, views, etc.), and we have every intention of making use of them to improve the stack (by making it leaner and more efficient, NOT by adding features willy-nilly!). ideally, you'll be able to pick and choose from the various AT features, using adapters to glue the functionality you need (and only what you need) into your own products. the first iterations of this will probably also have some warts. but please don't assume that plone/AT developers don't see the same problems that you see, and that they aren't willing and able to learn from their mistakes. -r ___ Zope maillist - Zope@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 )
Re: [Zope] VHost logs.
seb bacon wrote: >> Rob Miller wrote: >> >> In a manner of speaking, yes. That is, Apache needs to have correctly >> configured VirtualHost directives to handle the requests from the >> outside world appropriately, and Zope needs to have the SiteAccess >> product installed with correctly configured SiteRoots and access rules. >> It takes a little bit of time to figure out, but it works like a charm >> and is really quite simple, once you wrap your head around it. There's >> a great HOW-TO on this at http://www.zope.org/Members/anser/apache_zserver. > > > I would also point you to > > http://www.apache.org/docs/vhosts/mass.html > > for more info on the apache side, particularly this bit: > > The main disadvantage is that you cannot have a different log > file for each virtual host; however if you have very many virtual > hosts then doing this is dubious anyway because it eats file > descriptors. It is better to log to a pipe or a fifo and arrange > for the process at the other end to distribute the logs to the > customers (it can also accumulate statistics, etc.). This document refers to handling situations where you have a very large number of virtual hosts and don't want a separate VirtualHost directive for each one (because they're all very similar). This is not my case, but it could be the original poster's. You certainly CAN log to different log files if you have a VirtualHost directive for each host; I'm doing so. > > >> Another benefit of this setup is that it can allow for both regular HTTP >> and SSL connections to all of your sites, so you can remotely access the >> manage screens without sending your passwords in the clear. A HOW-TO >> for this lives at http://www.zope.org/Members/unfo/apache_zserver_ssl. >> I still haven't figured out a clean way to make it impossible to access >> sensitive areas UNLESS you're using SSL, however. Anyone out there >> doing this? > > > mod_rewrite is your friend. You just make a Rule that redirects > anyone accessing your site on port 80 to port 443, something like > this: > > > ServerName www.foobar.com > RewriteEngine on > RewriteRule ^/(.*) https://www.foobar.com/ > I don't want to force ALL access to port 443, only certain sections of the site which require authentication to access. I guess the above still holds true (your correction in a later message is noted), I just need to get more creative with my rewrite rules. Thanks for the tip. -rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] VHost logs.
Jason C. Leach wrote: > hi, > > With Apache in front of zope, do you disable Medusa? No. Apache merely acts as a proxy server. Apache gets a request, passes the request on to Zope, Medusa/Zserver serves up the results back to Apache, which then hands them out to the original requester. > And do you need to > set both Apache and Zope to have the vhosts (do I need to set up a vhost > twice?). In a manner of speaking, yes. That is, Apache needs to have correctly configured VirtualHost directives to handle the requests from the outside world appropriately, and Zope needs to have the SiteAccess product installed with correctly configured SiteRoots and access rules. It takes a little bit of time to figure out, but it works like a charm and is really quite simple, once you wrap your head around it. There's a great HOW-TO on this at http://www.zope.org/Members/anser/apache_zserver. Another benefit of this setup is that it can allow for both regular HTTP and SSL connections to all of your sites, so you can remotely access the manage screens without sending your passwords in the clear. A HOW-TO for this lives at http://www.zope.org/Members/unfo/apache_zserver_ssl. I still haven't figured out a clean way to make it impossible to access sensitive areas UNLESS you're using SSL, however. Anyone out there doing this? -rob > > On Wed, 29 Nov 2000, Rob Miller wrote: > > >> Jason C. Leach wrote: >> >> >>> hi, >>> >>> Has anyone implemented there own logging for Virtual Sites? >>> >>> I was thinking in the site rules an External Method could be called, >>> passed the Request obj, and from that generate logs for virtual sites. >>> >>> If anyone has done that, or knows of a better way I'd be interested in >>> hearing it. >> >> I'm not exactly sure what your goal is... but if you're just trying to >> generate separate log files for each of your virtual sites, I'm accomplishing >> that by having my virtual sites proxied behind Apache (using SiteAccess and >> the ProxyPass directive) and specifying the log files for each VirtualHost in >> my httpd.conf file. You can control the contents of the logs using the >> LogFormat directive; I use the same format for all of my virtual hosts, but I >> think you can specify a different log format for each host if you desire. >> >> If you're trying to accomplish something else which requires you to handle the >> logging on the Zope side of things, then it seems to me that your idea would >> work, although I'd want to find out how my server performance would be affected... >> >> Hope this is useful for you, >> >> rob >> >> ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] VHost logs.
Jason C. Leach wrote: > hi, > > Has anyone implemented there own logging for Virtual Sites? > > I was thinking in the site rules an External Method could be called, > passed the Request obj, and from that generate logs for virtual sites. > > If anyone has done that, or knows of a better way I'd be interested in > hearing it. I'm not exactly sure what your goal is... but if you're just trying to generate separate log files for each of your virtual sites, I'm accomplishing that by having my virtual sites proxied behind Apache (using SiteAccess and the ProxyPass directive) and specifying the log files for each VirtualHost in my httpd.conf file. You can control the contents of the logs using the LogFormat directive; I use the same format for all of my virtual hosts, but I think you can specify a different log format for each host if you desire. If you're trying to accomplish something else which requires you to handle the logging on the Zope side of things, then it seems to me that your idea would work, although I'd want to find out how my server performance would be affected... Hope this is useful for you, rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
[Zope] Second Try: Zope FTP tunneled through SSH1
Nobody picked up at all on my first post of this question, so I'm hoping I have better luck this time. Does anyone out there even have any leads for me in this situation? Here's the repost: I'm trying to tunnel FTP connections to a Zope server through SSH1 to ensure that passwords don't get sent in the clear. (I know the session will still be unencrypted; that's okay for now.) I've followed the instructions for doing so that are available at http://www.employees.org/~satch/ssh/faq/ssh-faq-5.html#ss5.6, and doing so I can successfully create an FTP session through an SSH tunnel. The problem is, once I've logged in, Zope won't let me do anything. It responds to every valid command with a single word: 'Unauthorized.' I know I'm connected to the server, because 'cd'ing into a nonexistent directory gives me a 'No Such Directory' response, byt 'cd'ing into a valid directory gives me th 'Unauthorized' response. If I connect directly to the FTP server using the same login and password, all works well. If this can't work, does anyone have Zope FTP server connections being tunneled through an encryption layer? I'd really like my developers to be able to use their favorite editor with an FTP connection, but I really don't want management-capable passwords flying around in the clear. Thanks for your time and assistance, rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
[Zope] Zope FTP through SSH1
Hi, I'm trying to tunnel FTP connections to a Zope server through SSH1 to ensure that passwords don't get sent in the clear. (I know the session will still be unencrypted; that's okay for now.) I've followed the instructions for doing so that are available at http://www.employees.org/~satch/ssh/faq/ssh-faq-5.html#ss5.6, and doing so I can successfully create an FTP session through an SSH tunnel. The problem is, once I've logged in, Zope won't let me do anything. It responds to every valid command with a single word: 'Unauthorized.' If I connect directly to the FTP server using the same login and password, all works well. Anyone here have any ideas why this might be? Thanks for your time and assistance, rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] secure management with Site Access?
David Elkins writes: > Have you tried setting up something like zope.domain.com with the https > SiteRoot and Apache pointing directly at manage? Pointing a virtual host directly at manage I can do, but where do I put the SiteRoot? If I put an https SiteRoot in a directory, then plain ol' unencrypted access won't be happening... all of the URLs inside the site will point to 'https://' instead of 'http://', right? Or am I missing something? > You could additional add > in a rewrite/proxy command to redirect to the SSL side for > www.domain.com/manage. Again, getting Apache to do the right thing I've got figured out. It's just the SiteRoot causing the problem; all of the links inside the folder with the SiteRoot will be either 'http' or 'https', but I'm looking for some solution that will allow either, depending on the previous REQUEST. I know that SiteRoot is basically a special kind of access rule, maybe a smarter access rule can be used instead of a SiteRoot... Thanks for your feedback, and please let me know if there's something that I'm not seeing. rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
[Zope] secure management with Site Access?
Is anyone using the Apache/mod_proxy/SiteAccess method of using Zope with the ability to access a given virtual host both with and without SSL encryption? If so, can you give me some pointers on how to accomplish this? I'd like to use SSL for all of my management activity but have the main site be available for non-encrypted viewing. The 'SiteRoot', though, will only rewrite the URLs with either 'http' or 'https', and isn't smart enough to do one or the other depending on the form of the original request. Is what I want to do even possible, or will I have to resort to using some sort of PCGI solution to accomplish my goal? thanks, rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] "REQUEST" a string object?
On Tue, 25 Jul 2000, Dieter Maurer wrote: > Rob Miller writes: > > > > > > > > > > > > > There are "sequence-index" and "sequence-number" variables > defined by "dtml-in". Thus, you need not count yourself. > Note: use as "_['sequence-*']" inside "...". I don't think this works for me though, because I don't want to count all of the sequence items, just the ones that match a certain criterion. Specifically, I'm stepping through all of the files in a directory, but I'm only displaying (and thus counting) the ones that are themselves directories. Is there a way to leverage the "sequence-" variables to help me here? > > > The error I get, running Zope in debug mode, is this: > > > >Error Type: AttributeError > >Error Value: 'string' object has no attribute 'set' > Is is possible that some of your folders has > a REQUEST property or acquires it? No, that's not the case. Nice thought, though. I've discovered (with Jonothan Farr's help) that the problem doesn't occur in Zope 2.1.6, only in 2.2.0. This leads me to think that a) it's a new bug in 2.2.0 I've uncovered or b) there's something different about the way Zope 2.2.0 handles the REQUEST object that I need to understand and account for. I've discovered (quite by accident) that an attempt to add a Squishdot installation (Squishdot-0-4-1) into Zope 2.2.0 generates the same error that my code generates. There's a note on the Squishdot page that indicates that they know it doesn't yet work with 2.2.0, and that it will soon be resolved (by Squishdot-0-4-4, they promise). There's no indication, though, of whether the Squishdot author(s) know what the problem is at this point. I'm considering making a post to Zope-dev describing the problem and asking if anyone knows whether this is a feature or a bug that I'm bumping into. rob (p.s. I've just had a thought that I might be able to force the loop to sort the files by type, which would at least guarantee that all of the directories would be consecutive. I could also maybe construct a new list of just the directories and then iterate over the new list. Either of these may allow me to use the "sequence-" values constructively. I want to learn what's going on here, though, and I'd rather not have to resort to complicating my algorithms. Counting stuff is a basic programming task, it should be able to be accomplished in a straightforward manner.) ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] Your feedback: what should DateTime strftime() behavior be?
+1 On Tue, 25 Jul 2000, Brian Lloyd wrote: > Hi all - > > There has recently been some confusion over the expected > behavior of various approaches to DateTime formatting in > Zope regarding timezone representation. I would like to > resolve this for the next release by making a proposal > and asking you to reply to the list with a "vote": > > +1 == agree > > +/-0 == no strong opinion > > -1 == disagree > > > So then, here is the situation. In Zope 2.2 (and earlier), > formatting a date using either: > > > > > > ...would give you the date *formatted based on GMT rather than > uthe timezone (usually local) representation of the object*. > Simply doing: > > > > ...however, would print the date in the current timezone of > tthe datetime object. > > Many feel that this difference is unintuitive and a pain. The > proposal is that both: > > > > > > ...would be changed to apply the format to the current TZ > rrepresentation of the object rather than convert to GMT. Of > course, this could be a problem if there are people currently > counting on the output being GMT, which is why we're putting it > to a vote. If this change is made for 2.2.1, those who still > wanted the output in GMT could just call the 'toZone()' method > of the datetime object to get a GMT version before formatting: > > > > > What do you think? > > > Brian Lloyd[EMAIL PROTECTED] > Software Engineer 540.371.6909 > Digital Creations http://www.digicool.com > > > > ___ > Zope maillist - [EMAIL PROTECTED] > http://lists.zope.org/mailman/listinfo/zope > ** No cross posts or HTML encoding! ** > (Related lists - > http://lists.zope.org/mailman/listinfo/zope-announce > http://lists.zope.org/mailman/listinfo/zope-dev ) ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] "REQUEST" a string object?
On Mon, 24 Jul 2000, Jonothan Farr wrote: > I couldn't reproduce this. The following code works for me in Zope 2.1.6, > verbatim. Ah-ha! It was on 2.2.0 that this code failed. And.. sure enough, a quick check on the Zope 2.1.6 install I still have around shows that the code works there for me as well. Thank goodness; I was pulling my hair out trying to figure out what I was doing wrong, but it seems likely to be a bug in the new release that was causing my problem. > > > > > > > > > > > > > > > > > > > In case we end in the middle of a row... > > > > > > > Maybe you can play with this as a starting point. > > I added a couple of REQUEST.set() calls at the top to simulate variables passed > from a form, I assume. > > 'local' is my LocalFS object. Are you using this code in a statement > or are you actually serving it as a .dtml file from the local file system? It may be academic now, but I'm using it in a .dtml file that is called as a method by a python product. Actually, what I'm doing is hacking LocalFS to be an image gallery, so that all I have to do is drop a bunch of image files in a certain directory on my hard drive and they'll show up on the www all thumbnailed and prettified, via the magic of Zope. The code that you saw is from my replacement for the 'methodBrowse.dtml' file in the LocalFS product. I know that there are some other PhotoAlbum type products out there that do something similar, but I wanted the exercise. Thanks for your help... I guess now I report this as a bug. I've seen some mention of a "Collector" for this sort of thing; I'll go dig around zope.org (as soon as it comes back up) to find out what that is and how to use it. Unless of course any Zen masters notice this thread and decide to tackle the problem straightaway... ;-] rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
[Zope] "REQUEST" a string object?
Arrggh! I'm getting very frustrated trying to accomplish something that should be easy. I'm stepping through all of the files in a LocalFS directory, and I want to dynamically create a that displays all of the sub-directories of the current directory. I want to inject a "" every so often, so I need to count the number of directories I've found so far to see whether I need to start a new row or not. I've spent a few hours digging through the list archives, so I know that creating a manually managed counter variable is not a straightforward affair. Here's the code that I have: In case we end in the middle of a row... The error I get, running Zope in debug mode, is this: Error Type: AttributeError Error Value: 'string' object has no attribute 'set' This is referring to the line in the middle of the loop, the one that actually does the incrementing. It thinks that REQUEST is a string, which of course doesn't have a "set" attribute. I've been banging my head on this all day and can't come up with an way to accomplish this trivial task. I don't want to push this into Python; this is simple UI code, not business logic, so it doesn't belong there. Besides, it's absurd to think that I would need to write an external method or some other Python method to do something as simple as this. (It may end up being true, but it's still absurd... ;-) ) Any help would be greatly appreciated. thanks for your time, rob ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Re: [Zope] sequence-item and ""
Diego Rodrigo Neufert writes: > Hi ppl... > > Why I cant use dtml-var "sequence-item">?? Because, as I understand it, anything within "" gets treated as Python code by the DTML interpreter. Thus "sequence-item" is parsed as an expression: sequence _minus_ item. A dash is not a valid variable name character in Python, but it is in DTML; this is unfortunate. > > Every time I try to access sequence-item under "" in a dtml-call dtml-var or > anything else I got this error: > > Error Type: NameError > Error Value: sequence > > Well, I found a solution: > > > > Now everything is ok... but I dont want to do this, I want to access the > *&@%$#@& sequence-item in "". I've searched through the list archives and come to the conclusion that your solution above is the cleanest way to handle this, for now. There are other ways, but they involve ugly-looking permutations of the "_" namespace variable, and they approach the splendor of Perl in their readability. The "right" solution, IMHO, would be to rename the "sequence-..." variables to a different set of names that doesn't cause the Python interpreter to choke. I seem to recall someone saying that this (or something similar) was being worked on, but, alas, for now is our best option. rob > > Can anyone help me? > > - > Diego Rodrigo Neufert > -webmaster > --- > (Magic Web Design) > (email) ([EMAIL PROTECTED]) > (curitiba) (pr) > > ___ > Zope maillist - [EMAIL PROTECTED] > http://lists.zope.org/mailman/listinfo/zope > ** No cross posts or HTML encoding! ** > (Related lists - > http://lists.zope.org/mailman/listinfo/zope-announce > http://lists.zope.org/mailman/listinfo/zope-dev ) > ___ Zope maillist - [EMAIL PROTECTED] http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
[Zope] HTMLFile from within a method?
Greetings. I'm a fairly experienced Python programmer, but am new to Zope.
I'm working on a Python product, and through this process I'm learning my way
around the Zope innards. I'm running Zope 2.2.0b3 on an RH6.1 box.
By declaring a class member that is an instance of the HTMLFile class, I
can publish dtml files. This is right out of the "Boring Product" How-To; it
looks like this:
--
class Product(Implicit, Persistent, RoleManager, Folder):
index_html = HTMLFile('index', globals())
--
Then when I browse to an installed instance of my Product, I'll see the
appropriate output generated by my 'index.dtml' file, which is in the same
directory as my Product's python modules.
What I'm having trouble doing is something similar but from within one of my
Product class's methods. I've tried this:
--
class Product(Implicit, Persistent, RoleManager, Folder):
def form_handler(self, REQUEST=None):
return HTMLFile('form_results', globals())
--
This causes the output to be munged, like so:
... and so on ...
I've also tried:
--
class Product(Implicit, Persistent, RoleManager, Folder):
form_results = HTMLFile('form_results', globals())
def form_handler(self, REQUEST=None):
return self.form_results()
--
This causes zope to attempt to publish the object, but the object (my DTML
method, stored in "form_results.dtml") doesn't seem to have access to the
Product's namespace. That is, I get an error like:
Error Type: KeyError
Error Value: standard_html_header
...even though there's definitely a standard_html_header defined.
Finally, I've tried this:
--
class Product(Implicit, Persistent, RoleManager, Folder):
def form_handler(self, REQUEST=None):
dtml = open('lib/python/Products/Product/form_results.dtml', 'r')
s = dtml.read()
dtml.close()
return s
---
This causes the right data to be passed out, but it's not treated
as dtml. A garbled version of my output appears in the browser window, and if
I view source I see:
... and so on ...
I know there are other ways to accomplish what I'm trying to accomplish, but I
really want to understand what's going on here. Can anyone enlighten me as to
why I'm seeing the results I'm seeing? Does anyone have suggestions for an
appropriately Zen approach to publishing dtml files from within a Product's
methods?
Thanks for your time,
rob
___
Zope maillist - [EMAIL PROTECTED]
http://lists.zope.org/mailman/listinfo/zope
** No cross posts or HTML encoding! **
(Related lists -
http://lists.zope.org/mailman/listinfo/zope-announce
http://lists.zope.org/mailman/listinfo/zope-dev )
