February 2000 - Zope-Dev

Re: [Zope-dev] Re. Scheduler product, anyone?
by Tres Seaver 05 Feb '00

05 Feb '00

Jerry.Spicklemire(a)IFLYATA.COM wrote > > Loren wrote: > > >... The crontab would run a Zope Dispatcher which would then dispatch > >Zope methods that had been scheduled from within Zope. Thus, Zope would > >acquire cronish capabilities. > > Mark Lesh (Mark, care to add anything?) developed an "event processor" > (pre-Zope) that had similar requirements. Since it had to run on both DOS > and Windows boxes it was done in Clipper, which has the very Pythonish > capacity of calling code "objects" stored as fields of a Clipper table! > > Zope and a standalone Python event loop can easily pass messages in the form > of records in a shared data store, e.g. a PostgreSQL database. > > > ... Do you think it's important to capture output? > > Not in itself, but it is certainly necessary to verify successful completion > of each event, and those events can involve creation of output. ZODB may or > may not be the target storage for such output, but an effective Event > Manager must at least be able to confirm the output's existence and > correctness, currency, etc. > > When an event fails an error event may be generated, whose first purpose is > to attempt to restart the failed event, including verifying and/or > reinitializing any prerequisites. An alert event (auto-page) may also be > generated, to notify the designated driver of a mishap. (Yuck!) > > We found that we needed to track : > > dependencies (conditions required prior to start, such as prior > events) > > normal runtime window (earliest start time, latest end time) > > normal runtime length (shortest expected completion period, longest) > > on-completion action (initiate subsequent event) > > on-error action (re-start event, alert event, etc.) > > >The kinds of scheduled tasks I'm thinking about are > > - email reminders > > - garbage collection > > - updating syndicated content from other sites > > > >Anybody have other examples with other use patterns? > > - promoting new or updated content through develop, test, production status > > - spamming users when new content gets published > > - notifying content maintainers of irate feedback (after spamming users) > > This sort of process may at first appear to fall outside Zope's target > problem domain, but applying Zope to the task of content managment leads > down this path quite directly. A Web Publishing Workflow process readily > maps to an event driven system, each event triggered by the completion (or > failure) of some prior event(s). > > Conceptually, the leap required to see Zope in this light is to think of > each request of an URL as a call to an object / method, not just a pointer > to a displayable Web "page". Jon Udel's Keynote from the Zope Track points > straight to this sort of Distributed Web Object model. > > see : http://udell.roninhouse.com/ZopeKeynote.html I have been thinking about what it would take to add "Events" to Zope, in the vein of the Property/Event/Method model of VB/Delphi/JavaBeans. The "simplest thing that could possibly work" seems to me to inject the "Obseerver" design pattern into Zope's current PropertyManager and ObjectManager classes: * PropertyManager would have installable "event handlers": - 'BeforePropertyUpdate': called just before modifying a property value, passing the object, the property name, the old value, and the new value. This handler could do anything (send mail, write to an external data store, whatever), and would return true if the change was OK (this is actually "observer-with-veto") - 'AfterPropertyUpdate': called after modifying a property value, passing the object, the property name, and the new value (this is a "pure" observer). * ObjectManager would have a set of similar handlers: - 'BeforeChildAdd' - 'BeforeChildDelete' - 'BeforeChildReplace' Something like this approach seems necessary to me to allow for generalize workflow applications in Zope. How workable does this scheme look? Tres. -- ========================================================= Tres Seaver tseaver(a)palladion.com 713-523-6582 Palladion Software http://www.palladion.com

1 0

tracking user usage
by ethan mindlace fremen 05 Feb '00

05 Feb '00

A long time ago in a galaxy far, far, away, a Zopista said that I could easily track user's space utilization. Could someone point me to the hooks or wherever else I need to go to implement something like this? Specifically, I just need to track how big the stuff in a SiteAccess folder gets. (of course, i have to cross-reference that with disk quotas, but that's another kettle of fish) ~ethan fremen (who wishes he could just #include another data.fs from the "parent" data.fs)

1 0

Re: [Zope-dev] Re. Scheduler product, anyone?
by Anthony Baxter 05 Feb '00

05 Feb '00

>>> "Loren Stafford" wrote > The kinds of scheduled tasks I'm thinking about are > - email reminders > - garbage collection > - updating syndicated content from other sites - checking for automated workflow tasks who's preconditions have been met, and executing them. > Because I'm also thinking about repetitive tasks, I'm not anxious to have to > capture the outputs of the scheduled methods. I'd have to add them to the > end of a log file, and have to worry about what happens when more than one > copy of the method is running. I thought I'd expect scheduled methods to > have side-effects, but no STDOUT output, and throw away any output that > might be created. Do you think it's important to capture output? Absolutely. A log file would be _bare_ minimum - ideally, some sort of pluggable logging architecture could be used so that you could slot in a smarter logger. Anthony

1 0

RE: [Zope-dev] ZServer(FCGI) stability
by Amos Latteier 04 Feb '00

04 Feb '00

> Unfortunately our site which runs both under straight ZServer > and through > FastCGI hangs and requires restarting every day. Here is some debuging > info (Zope-2.1.3). I believe that this problem is fixed in the latest version of FCGIServer.py in CVS. If you can't get it from CVS let me know and I'll mail it to you. Please let me know if the latest version does not solve your problem. Thanks! -Amos -- Amos Latteier mailto:amos@digicool.com Digital Creations http://www.digicool.com

2 2

RE: [Zope-dev] Re. Scheduler product, anyone?
by Jerry.Spicklemire＠IFLYATA.COM 04 Feb '00

04 Feb '00

Loren wrote: >... The crontab would run a Zope Dispatcher which would then dispatch >Zope methods that had been scheduled from within Zope. Thus, Zope would >acquire cronish capabilities. Mark Lesh (Mark, care to add anything?) developed an "event processor" (pre-Zope) that had similar requirements. Since it had to run on both DOS and Windows boxes it was done in Clipper, which has the very Pythonish capacity of calling code "objects" stored as fields of a Clipper table! Zope and a standalone Python event loop can easily pass messages in the form of records in a shared data store, e.g. a PostgreSQL database. > ... Do you think it's important to capture output? Not in itself, but it is certainly necessary to verify successful completion of each event, and those events can involve creation of output. ZODB may or may not be the target storage for such output, but an effective Event Manager must at least be able to confirm the output's existence and correctness, currency, etc. When an event fails an error event may be generated, whose first purpose is to attempt to restart the failed event, including verifying and/or reinitializing any prerequisites. An alert event (auto-page) may also be generated, to notify the designated driver of a mishap. (Yuck!) We found that we needed to track : dependencies (conditions required prior to start, such as prior events) normal runtime window (earliest start time, latest end time) normal runtime length (shortest expected completion period, longest) on-completion action (initiate subsequent event) on-error action (re-start event, alert event, etc.) >The kinds of scheduled tasks I'm thinking about are > - email reminders > - garbage collection > - updating syndicated content from other sites > >Anybody have other examples with other use patterns? - promoting new or updated content through develop, test, production status - spamming users when new content gets published - notifying content maintainers of irate feedback (after spamming users) This sort of process may at first appear to fall outside Zope's target problem domain, but applying Zope to the task of content managment leads down this path quite directly. A Web Publishing Workflow process readily maps to an event driven system, each event triggered by the completion (or failure) of some prior event(s). Conceptually, the leap required to see Zope in this light is to think of each request of an URL as a call to an object / method, not just a pointer to a displayable Web "page". Jon Udel's Keynote from the Zope Track points straight to this sort of Distributed Web Object model. see : http://udell.roninhouse.com/ZopeKeynote.html Later, Jerry S.

1 0

ZServer(FCGI) stability
by Pavlos Christoforou 04 Feb '00

04 Feb '00

Hello - Unfortunately our site which runs both under straight ZServer and through FastCGI hangs and requires restarting every day. Here is some debuging info (Zope-2.1.3). Unhandled exception in thread: Traceback (innermost last): File "/usr/local/Zope-2.1.3-src/ZServer/PubCore/ZServerPublisher.py", line 97, in __init__ response._finish() File "/usr/local/Zope-2.1.3-src/ZServer/FCGIServer.py", line 699, in _finish 'log_request')) File "/usr/local/Zope-2.1.3-src/ZServer/FCGIServer.py", line 564, in push if send: self.initiate_send() File "/usr/local/Zope-2.1.3-src/ZServer/medusa/asynchat.py", line 206, in initiate_send self.refill_buffer() File "/usr/local/Zope-2.1.3-src/ZServer/medusa/asynchat.py", line 190, in refill_buffer self.producer_fifo.pop() File "/usr/local/Zope-2.1.3-src/ZServer/medusa/asynchat.py", line 264, in pop del self.list[0] IndexError: list assignment index out of range I suppose I could modify the fifo class in asynchat to check for empty self.list before attempting to index it, but I believe such a situation should not normally arise (ie calling del self.list[0] on an empty list). Seems to me that the problem is with FCGIServer.py. As I need a temporary solution *quickly* I will modify the fifo class and hopefully I won't create more problems ... Any ideas? Pavlos

1 0

Fixing Data.bbb
by Hannu Krosing 04 Feb '00

04 Feb '00

Hi. I'm currently in a quite embarrassing situation. A site has been running for about half a year non-stop but now a problem arise when I tried to pack the database : there is a corrupted record dated 08-08-1999 which makes it impossible to start up the database. I was able to get into a state preceeding that date by running fixbbbtt.py, but a lot of work has been done since, and the site should even look absolutely different by now. Is there an easy way to fix things, or is the only possibility for me becoming an expert in ZODB2 format and doing it by hand ? ------------- Hannu

2 3

FastCGI protocol error
by news-list.zope.dev＠innominate.de 04 Feb '00

04 Feb '00

hi, i'm using zope-2.1.1 with apache-1.3.9 and mod_fast/snap_oct using a socket to communicate with the zserver. once in a while (every 100th request or so) i get this problem here (see log files below). and the very same happens with zope-2.1.3 on my machine at home. OS is linux-2.2.14... i'm willing to do some debugging myself (later this afternoon or during the weekend) but i wanted to know if others have this problem as well or have any idea... cheers thilo -----------------------------[ apache's error_log ]------------------------ [Fri Feb 4 10:15:29 2000] [error] [client 10.0.0.25] FastCGI: comm with server "/var/httpd/html/Zope" aborted: protocol error: invalid version: 61 != FCGI_VERSION(1) -----------------------------[ zserver's log ]----------------------------- 2000-02-04T09:15:29 ERROR(200) ZServer uncaptured python exception, closing channel <__repr__ (self) failed for object at 87b47b0 (addr='\377\277\001')> (socket.error:(32, 'Broken pipe') [/usr/local/lib/Zope-2.1.1-src/ZServer/medusa/asynchat.py|initiate_send|211] [/usr/local/lib/Zope-2.1.1-src/ZServer/medusa/asyncore.py|send|237]) -------------------------------[ apache's returns this ]------------------- HTTP/1.1 200 OK Date: Fri, 04 Feb 2000 09:15:28 GMT Server: Apache/1.3.9 (Unix) (Red Hat/Linux) mod_fastcgi/2.2.3 X-Powered-By: Zope (www.zope.org) Python (www.python.org) Content-Length: 12011 Connection: close Content-Type: text/html <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"> <HTML><HEAD> <TITLE>200 OK</TITLE> </HEAD><BODY> <H1>OK</H1> The server encountered an internal error or misconfiguration and was unable to complete your request.<P> Please contact the server administrator, root@localhost and inform them of the time the error occurred, and anything you might have done that may have caused the error.<P> More information about this error may be available in the server error log.<P> </BODY></HTML> ---------------------------------------------------------------------------- -- mezger(a)innominate.de innominate AG networking people fon: +49.30.308806-11 fax: -77 web: http://innominate.de pgp: /pgp/tm

1 0

ODBC on linux
by Gabor Melis 04 Feb '00

04 Feb '00

Hi gentlemen Is there en effort to bring ODBC drivers to Zope on linux? It seems to me that we have iodbc and unixodbc. Are these stable enough to work with? On a slightly different note, is there a way to connect to an oracle db on another computer *without* having any oracle specific libraries? Cheers, Gabor Melis

4 3

Catalog Bug still in 2.1.3
by Jason Spisak 03 Feb '00

03 Feb '00

Michel, I just uploaded 2.1.3 and noticed the bug fixes you had commited weren't in that release. Do you know when those will be included in an official Zope release? All my best, -- Jason Spisak webmaster(a)hiretechs.com

1 0