April 2001 - Zope-Dev

Optimizing Zope
by Stefan Bambach 03 Apr '01

03 Apr '01

Hello, Running Zope 2.2.2 on Dual PIII 933 with 1GB of RAM. This machine is not fast enough to handle my requests (about 800 Users using this system). It's a very complex document managing system with about 5000 documents actually and growing every day. Search requests are with Python backend. The backend sends lists of objects with methods and data. I print the data in DTML sequences calling the objects methods. System is a 2 server system, which do the data syncing over xmlrpc. One system is a Windows NT and the other is Linux. The high load results on Windows NT in Dr. Watson system crashes and on Linux only zope goes down. Windows NT crashes very often (about every hour), while Linux is crashing every few month. 1.) I tried to optimize it with simple objects, that only have the data in it and then print it with DTML (<dtml-var "variablename">). This is faster, because of the simpler objects. 2.) I tried running zope with apache's cgi, fastcgi over sockets, fastcgi over TCP/IP. Result: Slower than only Zope. 3.) Is there a way of avoiding error 'competing conflicts' in the internal Zope Classes ? This crashes the NT box very often. So the admin have to restart the service all the time :-( And it is not sure that it comes up again. Sometimes a reboot is the only way. Any Idea ? It doesn't matter how stupid it sounds. ciao. Stefan

2 1

User Authorization via IP (Domains) Bug in Zope 2.3.x
by Stefan Bambach 03 Apr '01

03 Apr '01

Hello, I used some Domain-Based users in Zope 2.2.2. After upgrading to Zope 2.3.0 and 2.3.1 these users don't work anymore. - Is it a bug ? - Has the syntax been changed in any way ? ciao. Stefan

1 0

New fishbowl proposal: PathIndexes
by Tres Seaver 03 Apr '01

03 Apr '01

Please review and comment: <http://dev.zope.org/Wikis/DevSite/Proposals/PathIndexes> Tres. -- =============================================================== Tres Seaver tseaver(a)digicool.com Digital Creations "Zope Dealers" http://www.zope.org

1 0

webdav GET method
by Terry Kerr 03 Apr '01

03 Apr '01

Hi, I find webdav useless in editing zope content as GETting a DTMLMethod for example, returns the processed method rather than the source. This is expected since webdav extensions to HTTP do not redefine the HTTP GET method, hence a webdav GET is simply a HTTP GET. I had a quick search of the zope-dev archives and some webdav resource, but surprisingly didn't find an discussion on this problem. What are peoples thoughts on this? Am I missing something? terry -- Terry Kerr (terry(a)adroit.net) Adroit Internet Solutions (www.adroit.net) Phone: +61 3 9563 4461 Fax: +61 3 9563 3856 ICQ: 79303381

2 1

SkinScript Questions
by Michael R. Bernstein 03 Apr '01

03 Apr '01

Hi all, I had some questions about SkinScripts: 1) Are there any naming conventions for SkinScripts? 2) Is there any reason (other than factoring) to split SkinScripts up, or is putting all your declarations in one SkinScript ok? Thanks, Michael Bernstein.

3 2

Zope 2.3.1-compatible ZEO Release
by Chris McDonough 02 Apr '01

02 Apr '01

There has been a new ZEO (Zope Enterprise Object) release, primarily made to achieve compatibility with code in Zope 2.3.1. If you are running ZEO, you need upgrade to ZEO release 0.5.0 before you can use Zope 2.3.1 reliably. The new release also fixes bugs found in previous releases, and can be used with Zopes prior to 2.3.1. For more information, see http://www.zope.org/Products/ZEO.

1 0

INSTANCE_HOME and Win32
by Andy McKay 02 Apr '01

02 Apr '01

In 4am's how to on using INSTANCE HOME he says "copy start and stop to the site directory". Whats start and stop? My second Zope instance is wanting to lock the same file on start up... E:\Zope\lib\python\ZODB\lock_file.py line 115 in lock_file -- Andy McKay.

2 2

Can't _setObject???
by Lalo Martins 02 Apr '01

02 Apr '01

Hi there I just found out (while coding ZUnit 0.2) that ZUnit 0.1 doesn't work on Zope 2.3.x for some reason. The method manage_startTests of the TestRunner class creates an instance of TestResults, runs the tests with it (using RESPONSE.write(), if this makes a difference), then adds it to itself using self._setObject(). On Zope 2.2.x this works like a charm. On 2.3.x it doesn't bomb, doesn't raise an exception and doesn't set my house on fire, but the TestResults object simply isn't persisted. We tried all kinds of desperate things, like setting _p_changed on both TestRunner and TestResults, and doing a get_transaction().commit() - nothing changes. Out of curiosity, we enabled the addlist in TestRunner and sure enough, you can add objects via the addlist and they persist normally. Anyone has a clue? Of all things that changed between 2.2 and 2.3, which one can be killing us like that? []s, |alo +---- -- I say a prayer now our love's departed That you'll come back to stay Bring back the perfect day http://www.laranja.org/ mailto:lalo@laranja.org pgp key: http://www.laranja.org/pessoal/pgp Brazil of Darkness (RPG) --- http://www.BroDar.org/

1 0

Medusa and HTTP/1.1
by Seb Bacon 02 Apr '01

02 Apr '01

I've been yelling at my ISP (BTopenworld) for having a badly configured transparent proxy cache, which caches all content *unless* the source specifies otherwise. It's been really messing up my development, and of course it's a concern for my users too. Hopefully they'll sort it out. Anyway, while I was researching for ammunition, I read all the HTTP/1.1 specs (RFC2068). Interestingly although the RFC strongly recommends making unvalidated objects uncacheable, it stops short of forbidding it. I wondered if it might be a good idea to make the ZServer more HTTP/1.1 compliant, by always adding validators to all objects (at present validators are only added to image / file objects). This is the recommended behaviour for servers. In the case of the ZServer, I guess the correct behaviour would be to have every page object return a Last-modified header, which defaults to the modified date of the newest component of the page. This behaviour could then be modified using cache managers like the Accelerated HTTP Cache Manager. Just wondered what anyone else thought. seb

2 1

Re: [Zope-dev] Page Templates and nested templates
by Johan Carlsson 02 Apr '01

02 Apr '01

> There's a Tracker linked from the main ZPT Wiki page. You should be able to > add it there. On the other hand, I'd like to know how exactly you've tried > to nest macros, and how it failed for you. > > Cheers, > > Evan @ digicool & 4-am > Here is an examples: ID: here/master <html xmlns:tal="http://xml.zope.org/namespaces/tal" xmlns:metal="http://xml.zope.org/namespaces/metal" metal:define-macro="master"> <head metal:define-macro="head"> <title tal:content="here/title">sss title</title> <meta something=""> </head> <body bgcolor="#9999ff" metal:define-macro="body"> <div metal:define-macro="bodymacro"> <h1>This is a macro 21</h1> </div> <div metal:define-slot="bodySlot"> <h1>This is a H1 Header</h1> </div> </body> </html> Now notice this if I state use-macro for the METAL statement in the HTML-tag, all subsequent metal:define-macro statements gets replaced with the use-macro="here/master/macros/master>" ID: here/Page1 (version1) <html xmlns:tal="http://xml.zope.org/namespaces/tal" xmlns:metal="http://xml.zope.org/namespaces/metal"> <head metal:use-macro="here/master/macros/head"> <title tal:content="here/title">sss title</title> <meta something=""> </head> <body bgcolor="#9999ff" metal:use-macro="here/master/macros/body"> <div metal:use-macro="here/master/macros/body"> <h1>This is a macro 21</h1> </div> <div metal:define-slot="bodySlot"> <h1>This is a H1 Header</h1> </div> </body> </html> But if I instead apply the use-macro on the body and the head this are working perfectly, subsequent define-macros in the inner div gets turned in to use-macro="here/master/macros/body" though. ID: here/Page1 (version2) <html xmlns:tal="http://xml.zope.org/namespaces/tal" xmlns:metal="http://xml.zope.org/namespaces/metal"> <head metal:use-macro="here/master/macros/head"> <title tal:content="here/title">sss title</title> <meta something=""> </head> <body bgcolor="#9999ff" metal:use-macro="here/master/macros/body"> <div metal:use-macro="here/master/macros/body"> <h1>This is a macro 21</h1> </div> <div metal:define-slot="bodySlot"> <h1>This is a H1 Header</h1> </div> </body> </html> How I would want it to work is that if a higher level of macro is set to use-macro the subsequent level would remain unchanged (e.g. as define-macro) until I explicitly change it in to use-macro. I would then be able to set it to any macro definition, e.g. use-macro="here/master/macros/bodymacro" or use-macro="<some other path>/macros/<some other macro>" etc. I think what would make allot of sense and that's probably how it supposed to work, right? Best regards, Johan Carlsson

1 0