- Zope-Dev - Zope lists

Re: [Zope-dev] dtml-in batching
by Tres Seaver 24 Jan '01

24 Jan '01

richard(a)bizarsoftware.com.au wrote: > richard(a)bizarsoftware.com.au wrote: > > > > The dtml-in batching mechanisms are quite difficult to debug - the DTML > > documentation I have has examples which break it and there's no indication > > of why. It turns out the DT_In renderwb() code gobbles all exceptions from > > the rendering of the previous and next blocks. I've hacked my code so that > > there's no try/except clause any more. I'm not sure what exception it's > > trying to catch, but could it perhaps be made a little more picky? > > And here's a diff - I'm pretty sure this was the intended behaviour... Please post this as a "Bug w/ patch" to the Collector: http://classic.zope.org:8080/Collector That way we don't lose your fix in the list traffic (losing it along with our marbles is another problem :) Tres. -- =============================================================== Tres Seaver tseaver(a)digicool.com Digital Creations "Zope Dealers" http://www.zope.org

4 6

DTML block parsing
by Casey Duncan 24 Jan '01

24 Jan '01

I am working on a custom DTML tag (that I may unleash on the world when finished) that has brought to my attention an interesting feature of the DTML parser: It preparses everything in the block(s) between the start and end tags before the document containing the code is committed. My tag is a little different in that I don't want to put DTML inside it, therefore this parsing is undesirable. There is however one built-in tag that this is true for as well: <dtml-comment>. Many a thread has been burned about that issue. I propose a change to DT_String to allow a tag to turn off this preparsing feature. It should be easy enough to implement by adding a class attribute to the tag class such as disable_dtml_block_parsing = 1. Is this abhorrent to anyone? If not, I will make a patch for DT_String to support this. The beneficial side affect would be that <dtml-comment> could be "fixed" so that its contents need not be parsable anymore, just like every other language I know of. -- | Casey Duncan | Kaivo, Inc. | cduncan(a)kaivo.com `------------------>

3 5

Re: [Zope-dev] CSS file serving
by Dieter Maurer 24 Jan '01

24 Jan '01

richard(a)bizarsoftware.com.au writes: > > http://www.zope.org/Members/haqa/ZStyleSheet > > Hrm - that's a little heavy for what I want - which is just an analog of > HTMLFile but for CSS files. That is, something I can add to my Python > product as an attribute. Definitely no management or other sorts of > trickiness required. Look at "App.ImageFile". If you are lucky, you can immediately use it to wrap your CSS files. If not, you should extend "OFS.content_types.guess_content_type" such that it recognizes the ".css" extension. Dieter

2 3

Membership and the 2.3 beta2 ... bug in the latter!
by Bill Anderson 23 Jan '01

23 Jan '01

Broke. broke badly. Not sur ecompletely why yet, but in 2.3b2, I can't move or copy a ZCatalog. There is one in the Membership interface, so as a result, no workie there. After deleteing int from teh membership interface folder, I get a traceback complaining about too many arguments in checkRoles (3 instead of 2). I haven't had time to dig into it yet, so I thought I would toss it up here so anyone with visibility to this issue can look at it. Bill Anderson

2 1

sql methods and the namespace stack
by Tim McLaughlin 23 Jan '01

23 Jan '01

Does anybody know why sql methods only search the request object instead of traversing the dtml namespace stack? TIA, Tim McLaughlin

2 1

Is manage_add for products broken?
by Jonothan Farr 23 Jan '01

23 Jan '01

Hello all, One of the Local File System users pointed out that calling manage_addLocalFS from dtml doesn't work. I did some checking and discovered that calling manage_addXXX doesn't work for any of the built-in products either. My testing was with 2.2.5. Am I crazy or didn't this work at one time? Is there a way to work around this? I couldn't find one. Thanks very much. Any help is appreciated. --jfarr

1 0

Specialist/Rack scalability
by Michael Bernstein 23 Jan '01

23 Jan '01

After comsidering the fedback I got from the previous 'Massive scalability thread, I decided to split my queries into two areas: Rack scalability and ZCatalog scalability. This email deals with the former. It seems clear that indexing and searching are more of a botleneck than storage/retreival. Nevertheless, so far I have not heard of anyone trying to store more than 60,000 objects in a rack. I need to know if there is any reason to suspect that storage (in the ZODB) or retreival performance would suffer if the number of objects was in the hundreds of thousands or even millions. Does anyone have anectodal or benchmark data that would suggest what happens with that many objects? Thanks, Michael Bernstein.

3 7

Re: [Zope-dev] How would I keep _v_* attributes from getting thrown away?
by Dieter Maurer 22 Jan '01

22 Jan '01

John D. Heintz writes: > I am using ZODB along with CORBA to expose peristent objects. Some of > my objects have per-connection/session state. > ... > My initial idea is to have a _v_position attribute ... Using "_v_" attributes for session state is a very bad idea. They are thread(!) local and not session local. This means, if a following request of your user (in the same session) happens to get assigned a different thread, it will read and update a different position. Chaos is garanteed (though you will see it only, if your site is loaded). Use a session product (-> zope.org) for session data. Dieter

2 1

DTML block parsing - Patch
by Casey Duncan 22 Jan '01

22 Jan '01

Ok here goes: Install these patches on a 2.3.0b1 installation. The files to patch are found in {Zope Dir}/lib/python/DocumentTemplate. Thus far it seems stable, and I would like for anyone willing to test it out. WHY SHOULD YOU? Although the original purpose of this patch is solve my own problem, it also solves a behavioral bug in <dtml-comment>. That is you no longer need correctly formed and balanced dtml code inside of a <dtml-comment></dtml-comment> pair. This allows them to function as you would expect. Code inside the comment tags is simply ignored. If you develop your own dtml tag class that you want to use to create blocks of something other than dtml, simply add the class attribute disable_dtml_block_parsing=1 along with blockContinuations=() in the tag's class definition. Here are the patches: *** DT_String.old Mon Jan 22 13:06:33 2001 --- DT_String.py Mon Jan 22 15:55:57 2001 *************** *** 272,277 **** --- 272,308 ---- sname=stag sstart=start sa=sargs + + if hasattr(scommand, 'blockContinuations') and\ + hasattr(scommand, 'disable_dtml_block_parsing'): + # Tag doesn't want the block parsed as dtml, so just find the + # end tag and don't parse what's inside + + while 1: + l=tagre.search(text,start) + if l < 0: self.parse_error('No closing tag', stag, text, sloc) + + try: + tag, args, command, coname= self._parseTag(tagre,scommand,sa) + except: + start = l + 1 # Skip malformed tags without raising exceptions + else: + start = l + len(tag) + + if not command and not coname: + # Block end tag + section = String(text[sstart:l], __name__=sname) + section._v_blocks = section.blocks = [section.raw] + section._v_cooked = None + blocks.append((tname, sargs, section)) + try: + r=scommand(blocks) + if hasattr(r,'simple_form'): r=r.simple_form + result.append(r) + except ParseError, m: self.parse_error(m[0],stag,text,l) + + return start + while 1: l=tagre.search(text,start) *** DT_Var.old Mon Jan 22 15:27:30 2001 --- DT_Var.py Mon Jan 22 15:28:14 2001 *************** *** 461,466 **** --- 461,467 ---- ''' name='comment' blockContinuations=() + disable_dtml_block_parsing = 1 def __init__(self, args, fmt=''): pass Feedback on this is definitely welcome. If it is generally well-liked, I will submit it to the collector. -- | Casey Duncan | Kaivo, Inc. | cduncan(a)kaivo.com `------------------>

1 0

dtml-in batching
by richard＠bizarsoftware.com.au 22 Jan '01

22 Jan '01

The dtml-in batching mechanisms are quite difficult to debug - the DTML documentation I have has examples which break it and there's no indication of why. It turns out the DT_In renderwb() code gobbles all exceptions from the rendering of the previous and next blocks. I've hacked my code so that there's no try/except clause any more. I'm not sure what exception it's trying to catch, but could it perhaps be made a little more picky? Richard -- Richard Jones richard(a)bizarsoftware.com.au Senior Software Developer, Bizar Software (www.bizarsoftware.com.au)

2 3