April 2000 - Zope-Dev

There's a new Wiki in town; DTMLWiki
by Evan Simpson 12 Apr '00

12 Apr '00

http://www.zope.org/Members/4am/DTMLWiki is a new Wiki on Zope.org for discussing DTML -- where it is, and where it should go. I've seeded it with three evolutionary proposals which we have been kicking around. Enjoy! Cheers, Evan @ digicool & 4-am

1 0

zope imap server?
by mindlace 11 Apr '00

11 Apr '00

Zopistas, Paul Everitt mentioned that someone was working on a Zope imap server. I'd very much like to get in touch with whoever's working on that, because it would be ideal for several projects I'm trying to put together. Mostly I'd like to use it as the core of a weblog/maillist software. Thanks, -- ethan mindlace fremen mindlace(a)imeme.net zope -&- imap email -&- mailing list weave your web with the web at http://imeme.net

2 1

product creation wierdness in 2.1.6
by Sam Gendler 11 Apr '00

11 Apr '00

I have been plagued by a very strange bug tonight. I am trying to create a base class for a zClass, and I had an typo on a line of the code, which caused an import to fail. No matter how hard I tried, I could not get get Zope to notice that the line had changed, despite many restarts, removals of .pyc files, etc. The exception would still get thrown, and the line of source quoted, always had source that no longer exited in the .py file. Finally, I thought that perhaps I should delete the product altogether, and then restart zope. Now the product does not show up at all. Ever. No matter what I do (including undo of the delete). Finally, I moved the source code out of zope's home directory, and blew zope away, and reinstalled from scratch from source. When I move my code back into lib/python/Products, it still fails to appear in teh product list, even as a broken product. I have looked at the source, and it does not look as though it has been modified, and this is an entirely new Data.fs file, so what could possibly be going on. Any ideas, anyone? This is not my first product (zClass or otherwise) and I have never seen this problem bfore. Zope is version 2.1.6 Please help. --sam

2 3

Fwd: Calling os.waitpid() From Secondary Thread Under Linux 2.2.12-20?
by Patrick Phalen 11 Apr '00

11 Apr '00

This just appeared on comp.lang.python. Can any Zope experts comment on whether this Linux threads comment has implications for Zope? ---------- ## Forwarded Message ## ---------- Subject: Calling os.waitpid() From Secondary Thread Under Linux 2.2.12-20? Date: Tue, 11 Apr 2000 17:48:11 GMT From: rwiserREMOVEthis(a)mindspring.com (Randy Wiser) I've written a small python program that spawns one child process from its main thread, and then creates a second thread. The sole purpose of the second thread (at this point) is to call os.waitpid(processID, 0) to wait for the child that the main thread created. This works fine under Solaris when I call os.waitpid() from the second thread. But, under Linux 2.2.12-20 I get the following OSError exception: [Errno 10] No child processes However, if (under Linux still) I call waitpid() from the main thread (as a test), the call does indeed wait until the child process terminates. After scouring some of the Linux newsgroups, I've gathered that threads spawned under Linux may have SIGCHLD signals (and other signals) blocked, and that this _may_ be the reason I get the error, even though the processID I pass is for a valid process. Unfortunately there seems to be no easy way to set the signal mask for the second thread. Has anyone else run into this situation, and if so can you advise how you solved it? -------------------------------------------------------

1 0

Problem in ZGadflyDA/__init__.py
by Gregor Hoffleit 11 Apr '00

11 Apr '00

There's a small but sometimes fatal problem in ZGadflyDA/__init__.py: __init__.py currently checks for the presence of a gadfly directory in Zope's var directory, and creates it if it's not yet there: j=os.path.join d=j(j(INSTANCE_HOME,'var'),'gadfly') if not os.path.exists(d): os.mkdir(d) os.mkdir(j(d,'demo')) Now if you start up z2.py as root, since __init__.py is called *before* z2.py tries to setuid/setgid, the gadfly directory will be created with the original uid and gid, i.e. root. Depending on the umask, this leaves the Zope process with insufficient permissions to access the gadfly directory. The suggested change is easy: Similar code is already in ZGadflyDA/db.py, so it should be safe to remove the lines above from ZGadflyDA/__init__.py. Gregor

2 1

incompatible time modules?
by Stefan Kuzminski 11 Apr '00

11 Apr '00

Hi all, I get the following error on starting zope.. Traceback (innermost last): File "/home/langur/Zope-2.1.6-src/z2.py", line 427, in ? import ZServer File "/home/langur/Zope-2.1.6-src/ZServer/__init__.py", line 123, in ? from HTTPServer import zhttp_server, zhttp_handler File "/home/langur/Zope-2.1.6-src/ZServer/HTTPServer.py", line 116, in ? from HTTPResponse import make_response File "/home/langur/Zope-2.1.6-src/ZServer/HTTPResponse.py", line 99, in ? from medusa import http_server, asyncore File "/home/langur/Zope-2.1.6-src/ZServer/medusa/http_server.py", line 746, in ? tz_for_log = compute_timezone_for_log() File "/home/langur/Zope-2.1.6-src/ZServer/medusa/http_server.py", line 729, in compute_timezone_for_log if time.daylight: AttributeError: daylight I built everyting from scratch on a linux box. The zope code seems to expect different things from the time module than the Python distribution includes. Is there a special zope version of the time module? I'm using python 1.5.2. Thanks, Stefan Kuzminski

1 0

Collector Status Update
by Tres Seaver 11 Apr '00

11 Apr '00

Working through the Collector, I have seen a need to separate out the patch for a bug from its description: for instance, the description is formatted using newline_to_br, which breaks any patch to Python code; patches need to be formatted inside <PRE> tags. I therefore added a shiny new "Patch" field, and will work on migrating those bugs/features with patches to move the patch. In the future, please stuff actual patches into this field (use 'diff -c' by preference). Tres. -- ========================================================= Tres Seaver tseaver(a)digicool.com tseaver(a)palladion.com

1 0

Re: [Zope-dev] TreeTag
by Tres Seaver 11 Apr '00

11 Apr '00

Chris Withers <chrisw(a)nipltd.com> > > <RANT> > Now that's the collector has been opened up you can see that certain > things have stagnated ;-) > If you search for TreeTag, you get seven items. > One is a patch fixing a problem I have (which is still there in 2.1.6), > another is a patch fixing a > problem I didn't know about, and one adds some really useful functionality. > They are all between 4 and 8 months old and yet all except one are still > 'pending' > Even with DC's limited resources that's not very good... > </RANT> > > There must be a better way, personally I'd like to know how I can help > and get these patches into the next point release as quickly as possible. > > Any ideas? As the guy who opened the Collector, and who is actively wading through it at the moment, I can sincerely say, "I feel your pain." I am working on Collector issues, prioritizing them for the next several releases -- patches as always get priority (assuming we can understand them, of course). We are actively working on ways to enable more community in the evaluate/integrate/test cycle, but are obviously not there yet. > PS: I was just about to go and fix these problems myself, so I'm very > glad the collector has opened up ;-) I'm glad that opening it up helped you. If you'd like to "vote" for a patch, try it out on a scratch Zope, with a copy of your "real" Data.fs, and verify that it doesn't raise any new problems, and that it fixes the reported problem. Send me mail, with the issue #'s in the subject, and let me know that it worked (or not -- and patches to patches are accepted, too :) Tres. P.S. I have two of the tree-tag patches (#913/1045) checked in already. -- ========================================================= Tres Seaver tseaver(a)digicool.com tseaver(a)palladion.com

2 2

Persistency of database method classes (pluggable brains)
by Dr. Peer Griebel 11 Apr '00

11 Apr '00

It's just me again... The ZSQL document mentions to define a database method class in a file in the Extensions directory. The ZCMG document says "you cannot define classes or objects in your External Methods and expect to be able to assign them to attributes of Zope objects." This is what I see in my application: I'm doing a SQL query. The result is stored in an array of object my own database method class. I save this array in an FSSession object to be used later on. This later usage results in problems. The array does not consist of objects of my own class (or of class r as defined in RDB.py) but of type extension class Record.Record. So how can I make my class definition persistent? The database method class lives in the Extension directory which means this class is not persistent by definition :-(( But perhaps I just overlooked something very simple. Still praying... Thank you! -- connection reset by Peer _______________________________________________________________________ Dr. Peer Griebel Tel. +49 7581 4831 23 Geschäftsführer Fax. +49 7581 4831 11 Knoll Informationssysteme GmbH http://www.knoll-is.de Dreiköniggasse 17 mailto:peer@knoll-is.de 88348 Saulgau privat: mailto:peer.griebel@gmx.de

1 0

Sorting/Reversing dtml-in with URL arguments
by Kevin Smith 11 Apr '00

11 Apr '00

Is it possible to set the sort order and reverse flags on a dtml-in tag using URL arguments or REQUEST variables? For example, sorting "employees" by "name" in index_html would be something like: index_html?sort for <dtml-in employees> ... </dtml-in> Kevin Smith Kevin.Smith(a)theMorgue.org carpe noctum

2 1