February 1999 - Zope

BoboPOS file locking
by Gregory Trubetskoy 21 Feb '99

21 Feb '99

Right now, if the database is locked by a process, no other process can (or should for fear of curruption) access it. I was wondering if there is any progress on making BoboPOS databases accessible from multiple processes? Is anyone working on addressing this at all? I know I've brought this up on this list before, but never quite got a definitve answer... :) Grisha Trubetskoy

3 2

Zope's Organization for Virtual Servers
by Jimmie Houchin 21 Feb '99

21 Feb '99

Hello, I am in the process of working with my webhost on the best way to provide Zope Hosting in a virtual server environment. My webhost like many is using Linux and Apache. I had a thought and a question I wanted to pose. What would it take to setup Zope in somewhat a similar many to the way Python is setup for cgi? What I mean by that is if Zope was setup in a specified location /usr/bin/Zope or something such. Then in their own directory they have the products, external methods, /var directory, pcgiinfofiles, etc. All the items they create and generate which are not a part of the Zope distribution. This way, when the website is accessed, pcgi would instantiate a new Zope instance (if necessary) for this specific site and populate it with the site's files. This would allow for the webhost to have a single Zope install which could be forked/instantiated for each site as required. This would also allow for simplified upgrading of Zope for the webhost and users. The webhost could install a new Zope in it's own directory, usr/bin/Zope_1.10.2 and the user would just have to update their pcgiinfofile to access the new Zope install. Possibly less things to break when updating. Wouldn't have to move/copy/backup the /var directory etc. simply to upgrade. This seams to be simpler on the end user and on the webhost sysadmin. Of course I may have an over-simplified view of things. :) Just a thought. Jimmie Houchin Now it's time to get into my new book. Learning GNU Emacs. :)

3 4

RE: [Zope] Shattered dreams... And glibc
by Shaw, Howard 20 Feb '99

20 Feb '99

Can anyone tell me whether the Zope precompiled binaries are statically linked or not? I'm running a Redhat 5.1 system, and I downloaded the most recent MySQL rpm, to see if it affected my Zope recurrent death problem at all. In the upgrade notes in the rpm, it mentions that Redhat 5.1 system's glibc-2.0.7-13 has a problem that causes some MySQL clients to die of memory problems. It recommends upgrading to glibc-2.0.7-9, which I am about to do. Since I compiled the MySQLmodule for Python directly from the ZMySQLDA package, on this system with a bad glibc, it seems probable that this might be my solution. I am going to download the upgraded glibc and recompile, and will inform the list of the result. Second question: The newest MySQL rpm, 3.22.16a-gamma-1, implies that some older client software using certain C calls to connect may need to be rewritten to work; and indeed, I cannot connect to the newer mysqld from Zope or PHPMyAdmin, although I can see with 'mysql' that it is running and does contain the same databases and same data. Any ideas on what I can do to make these two work together? This is not a priority issue, since my previous MySQL was installed in /usr/local from the precompiled statically linked tar.gz, and is still usable, so once I recompile MySQLDA with the newer glibc, I will downshift to the older MySQL... But I would like to know how to resolve the problem with the newer MySQL. Thanks for the help. Howard C. Shaw III St. Thomas High School Houston, Texas

2 1

Re: Zope Debian package?
by Gregor Hoffleit 20 Feb '99

20 Feb '99

Hi, since a few people asked: On Sat, Feb 13, 1999 at 11:19:16PM -0500, Christian Hudon wrote: > are you still working on a zope package for Debian? If yes, when is it > going to be out? I'd like to play with zope at the office, and so if you're > too busy I could build a zope package and upload it this week. at http://master.debian.org/~flight/zope/, there is a first try at packaging Zope-1.10pr1 for Debian (targetted for potato). The naming will change (1.10pr1>1.10 ;-), it has some rough edges, the docs are unfinished etc.pp., but it works well for me. zope-httpd is ZopeHTTPServer, readily configured to run via /etc/init.d (is this really a good idea ?). Expect to find a zserver package in the future as well. I'll announce when there is a real release to the greater public. Gregor

3 2

RE: [Zope] "proxy roles", search engine for documentation.
by Rob Page 20 Feb '99

20 Feb '99

Anthony: > Ok, more information needed: do the proxy roles augment or replace > the roles that the user currently has? Sorry for the delay in responding. They (proxy roles) _augment_ existing user roles (per Jim and a little micro-test I just did). Consider a folder: Container containing two documents: secret_doc and wrapper_doc wrapper_doc is:  <P>This is a wrapper. Below is the secret content:</P> <BR>   We can restrict View access to secret_doc to the role, secret_readers We can then give wrapper_doc a proxy role of secret_readers Obviously, as long as Anonymous can View wrapper_doc ANYONE can, transitively, see secret_doc. Let's say Bob is in the role guest_readers and NOT in secret_readers. If we restrict View access on wrapper_doc to guest_readers Bob will be able to view secret_doc through wrapper_doc but not directly. I know this is a little hard to follow (it is when I re-read it!) -- Best way is to understand all this is to play with it. --Rob (Zen 0.2) P.S. - We're at the thin-air limits of my comprehension so if your questions go much deeper I'll have to defer to the Zen 0.75+ club! :^)

1 0

Controlling visibility of dtml methods/documents -- a newbie question
by Butch Landingin 20 Feb '99

20 Feb '99

How do I stop direct access to a dtml method I've created in a folder? I don't mean specifying restricted access to it by modifying the security properties and specifying manager role access only -- this brings up a login screen -- what I want is an error screen that says that there is no such document. For a clearer example, the sample zope installation has the dtml methods standard_html_header and standard_html_footer defined. And the way these methods are used is that they are included in the index_html dtml method using <var> tags (i.e. they are not intended to be documents that are accessed directly). All this is clear to me -- the problem I have is that I can access those standard_html_header / footer methods directly -- by typing something like "http://mysite.com/standard_html_header" -- Is there any way of preventing this? One solution I tried was changing the security (BTW, I encountered some problems when trying to set values on this tab -- I kept getting errors -- has somebody reported this as a bug?) but what that did was restrict access to the dtml methods that used the standard_html_header (e.g. the index_html dtml method). I've read something about proxy roles and maybe this is the short term solution (I don't know -- I don't understand enough about it). But what I really want to do is to control visibility of an object -- something like a property I can switch on or off for each object so that when I try to access it directly (i.e. "http://mysite.com/folder/dtmlmethod" ) will display an error message saying that it doesn't exist -- but I can still access it by including it in my dtml code (e.g. "") and I'd also want it turned on for some dtml methods (e.g. for the index_html dtml method). Is there some Zope facility for doing this? _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com

2 1

ZServer (ImportError: No module named medusa)
by Oliver Thuns 20 Feb '99

20 Feb '99

Two questions: 1) Is it possible to use ZServer only to publish simple modules (ZPublisher) without the Zope stuff? 2) Why do I get this error? I only untared the archiv, no changes made. D:\thrir\opt\bobo\ZServer>python start.py Traceback (innermost last): File "start.py", line 17, in ? from medusa import resolver,logger,http_server,asyncore ImportError: No module named medusa Oliver

2 1

RE: [Zope] ZPublisher/ZTemplate questions
by Paul Everitt 20 Feb '99

20 Feb '99

James wrote: > My first question has to do with using templates with > ZPublisher: Will I > still get to use all the nifty "var", "expr" and flow control > constructs > in my dtml, or will I be limited because my stuff in not > running within > Zope? Do I need to get the ZTemplate package to do this, or does > ZPublisher have this stuff built in? You can use ZPublisher and ZTemplate. This is what Skip is doing at MusiCal with good results. > Second, where can I get documentation and examples on using > ZPublisher and > ZTemplate on their own? Uhh, that's a different point :^) http://www.zope.org/Documentation/Reference/ObjectPublishingIntro http://www.zope.org/Documentation/Reference/Trinkets > Third: Will ZPublisher run with the embedded Python/Apache > system, or do > I need to use pcgi? Can't comment on PyApache. You can use PCGI or you can do as Skip and Andreas do an use mod_proxy to connect Apache and ZPublisher. > Fourth: If I want to maintain state/session information > inside my code, > which of the above solutions will work better? I know that Apache > instances only last for N requests, does Python/Apache have a shared > memory system to handle this. (Sorry if this is off topic, > but I figure > someone here must have played with it) That's why you don't want to use PyApache. Have a separate ZPublisher long-running process. > Thanks for all the help, I'm really excited to get started > with Python. > ColdFusion is frustrating because it's so limited, and Java > Servlets are > frustrating because they are too much work. Oh you're just sweet talking us :^) --Paul

4 3

brain'ed dtml???
by Kent Polk 20 Feb '99

20 Feb '99

This was from a couple of weeks ago: > We've kicked around the idea of an ExternalFile object, and I > started a small project that fizzed out. My question was where to > draw the line, do we make it a simple pipe to one file, or a Folderoid > object that 'mounts' a part of a filesystem into Zope? The first > would be pretty simple, the second would be pretty hard, IMO. I wasn't ready to answer this the first time around. :^) What I am trying to do right now is to map URL entries built from a database query which builds an index table and include content located in the filesystem. Constraints are that the document URL appear to be a static URL (no GET query) and that I provide a unique DMTL wrapper for each folder which allows one to customize the wrapper that contains the filesystem-based data. Actually, I would like to obtain the included data from anywhere I need, not just from the filesystem. Now I know how to use pluggable brains from ZSQL to specify a class to get close to this, but what if the document isn't in the database and a query can't return it? I would like to have that same class behavior which would allow me to use a __getitem__ to use the 'filename' (which doesn't really exist anywhere) as info on how to include the content I need. I wrote a little external method - include_file() which takes the pathname from a folder attribute and the 'filename' which is called via a 'wrap' dtml method (wrap?filename) which includes the filedata in context. But instead of 'wrap/?filename' what I want is more like 'get/wrap/filename' where get sets the class behavior that knows how to find 'filename' and 'wrap' is the editable dtml in that folder which encapsulates it. Or I'd be very happy with 'wrap/filename', but I don't know how to assign __getitem__ behavior to a DTML document. :^) Alternately, I was thinking of writing a Product which allowed me to specify the dtml wrapper as an attribute and used a __getitem__ method to retrieve the filedata and encapsulate it in the wrapper. Suggestions? Thanks Kent

1 0

ZPublisher/ZTemplate questions
by James A. Hillyerd 19 Feb '99

19 Feb '99

Okay, so a couple of people have said that it looks like I want to use ZPublisher on it's own, instead of the entire Zope setup. It also looks like there are a number of other stubborn =) (read: we like things in files) people out there who are doing the same. My first question has to do with using templates with ZPublisher: Will I still get to use all the nifty "var", "expr" and flow control constructs in my dtml, or will I be limited because my stuff in not running within Zope? Do I need to get the ZTemplate package to do this, or does ZPublisher have this stuff built in? Second, where can I get documentation and examples on using ZPublisher and ZTemplate on their own? Third: Will ZPublisher run with the embedded Python/Apache system, or do I need to use pcgi? Fourth: If I want to maintain state/session information inside my code, which of the above solutions will work better? I know that Apache instances only last for N requests, does Python/Apache have a shared memory system to handle this. (Sorry if this is off topic, but I figure someone here must have played with it) Thanks for all the help, I'm really excited to get started with Python. ColdFusion is frustrating because it's so limited, and Java Servlets are frustrating because they are too much work. -james [] James A. Hillyerd <jamesh(a)altavista.net> Java Developer [] HyperGlyphics: http://www.hyperglyphics.com/ [] GPG Public Key Fingerprint for 1024D/9F956CDE (Expires 2000-02-01): [] C86F B073 92DF 1E24 EF0B 0118 6061 0FEC 9F95 6CDE

1 0