July 2000 - Zope-Dev

pdbtrack.el - track source file whenever runing pdb in an emacs buffer
by Ken Manheimer 16 Jul '00

16 Jul '00

I've put together an alternative, pared down approach to doing python debugging inside emacs - pdbtrack: http://www.zope.org/Wikis/klm/PDBTrack Once loaded, pdbtrack watches your (comint) process interaction buffers for the pdb stacktrace. When it sees it, and can find the file containing the current stacktrace focus, it pops the file up in a separate emacs buffer, centered on the current debug line. This way you continue to interact with pdb, stepping into and out of functions, moving up and down stack frames, anything that changes focus - and emacs displays the right place in the source in a companion buffer. Of course, pdbtrack works just fine with the zope interactive debugger (reachable, for example, via: % cd $SOFTWARE_HOME/lib/python % python >>> import Zope; app = Zope.app() >>> Zope.debug('Zope.app().whatever', d=1) [debug...] ) The big difference between pdbtrack.el and pdb.el is that you don't need to start your debugging session in a GUD (Grand Unified Debugger) buffer. I often have to arrange a lot of state in the interpreter before i can jump into debugging. This generally defeats switching to a GUD session, with its separate process - while pdbtrack just jumps in whenever i go into the python debugger, within the accumulated context. pdbtrack lacks a number of things that GUD does have. It doesn't highlight the current source line other than centering it in the source buffer, and it doesn't provide nicities like stepping and setting breakpoints from within the source buffer. However, pdbtrack is much much more lightweight than gud - and it might be interesting to see about providing some of the gud features, like debug keybindings in the source file buffers involved in the debug session, with this more open, decentralized model. In the meanwhile, pdbtrack goes a long way towards doing what i want - i hope it's useful to others. I've submitted an entry for vaults of parnassus, not sure how long it takes to show. To repeat from above, you can find it at: http://www.zope.org/Wikis/klm/PDBTrack Ken Manheimer klm(a)digicool.com

1 0

Re: [Zope-dev] Optimization and speed
by Stephan Richter 15 Jul '00

15 Jul '00

>It looks like the pages you are serving are mostly big tables. Both >Netscape and IE >won't reder anything unless they got everything in the table. > >I would try to time the 'wget' speed to verify if zope is slow or not. Well, I kinda did that using telnet. I telneted into port 80 and then asked for an HTML page. I took about 3 secs (that's okay) for Zope to render the HTML page and then it bursted it out all at once. As I read from the other E-mails, Zope does not stream HTML to the HTTP Server till it is done laying out the site. That is the real problem. But the problem is still not solved. I am still 50% slower than other sites, even though Zope does not stream the data out. I played with the SQL Method cache; maybe I have to increase it more. Regards, Stephan -- Stephan Richter CBU - Physics and Chemistry Web2k - Web Design/Development & Technical Project Management

1 0

Re: [Zope-dev] Optimization and speed
by Steve Alexander 15 Jul '00

15 Jul '00

Stephan Richter wrote: > > ><snip Method doc string> > > >Looks like it's more that just a tweak, though :-( > > So, how can I use this method? I do not understand the internals of the > ZServer completely. You've got a RESPONSE object. For HTTP stuff, it will be a HTTPResponse as defined in lib/python/ZPublisher/HTTPResponse.py. You can use the write(self, data) method of this response object to stream data directly to your client, independent (to a degree) of Zope's transaction processing. This is the method from HTTPResponse. def write(self,data): """\ Return data as a stream HTML data may be returned using a stream-oriented interface. This allows the browser to display partial results while computation of a response to proceed. The published object should first set any output headers or cookies on the response object. Note that published objects must not generate any errors after beginning stream-oriented output. """ if not self._wrote: self.outputBody() self._wrote=1 self.stdout.flush() self.stdout.write(data) So, you could set output headers and cookies as needed, then, before the rest of the processing on your page, have a <dtml-call "RESPONSE.write('the top of the page')">. Then again, I've never tried this. YMMV and all that. (*off to help cook... won't be reading email for a while*) -- Steve Alexander Software Engineer Cat-Box limited http://www.cat-box.net

1 0

RE: [Zope-dev] Announcing dev.zope.org
by Brian Lloyd 15 Jul '00

15 Jul '00

> > : http://dev.zope.org. > > The "Log in" link points to www.zope.org, which makes it > impossible to login to the new site (of course, not > "impossible" for someone who knows zope, who can just type > "dev" over "www" in the location box...) Thanks - I've made the necessary update so that people's authorization is valid over both www and dev (and other subdomains that may come later). Brian Lloyd brian(a)digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com

1 0

Security model idea from LoginManager: roles in a class-like hierarchy
by Lalo Martins 15 Jul '00

15 Jul '00

I think perhaps roles should take part in a class-like hierarchy. It would be useful to be able to say that, for example, "Author" is a subrole of "User", so that if I give "Author" for an user it automagically gets the permissions for "User". In my zope-coding experience this is a very realistic scenario for mass-community sites. >From the point of view of UI, this would mean each role would have to get a "homepage" in the folder it is defined, where you can control its set of "superroles". That shouldn't be too hard. I'll flesh out the proposal and post it to the appropriated wiki... but first I'd like to collect opinions from the list. (Of course, by its own nature, this feature is optional and backwards-compatible... if you don't want to give superroles to your roles, just don't.) []s, |alo +---- -- Hack and Roll ( http://www.hackandroll.org ) News for, uh, whatever it is that we are. http://zope.gf.com.br/lalo mailto:lalo@hackandroll.org pgp key: http://zope.gf.com.br/lalo/pessoal/pgp Brazil of Darkness (RPG) --- http://zope.gf.com.br/BroDar

1 0

passing namespace
by danchik 15 Jul '00

15 Jul '00

if I submit a form to some page that has frames, how do I pass the REQUEST namespace to those frames????

1 0

(no subject)
by danchik 15 Jul '00

15 Jul '00

if i call a page with some REQUEST variables (submited by form), how can I pass the namespace variables to the frames???

1 0

Making Folders callable
by Steve Alexander 14 Jul '00

14 Jul '00

Is there any reason that Folders aren't callable? That is, you can't directly render a folder with <dtml-var nameofafolder>. You have to do something like <dtml-var "nameofafolder.index_html(_.None, _)">. But, that's not quite the same because if the folder lacks an index_html, it gets acquired. Here's an extra method that can be added to lib/python/OFS/Folder.py, or to ObjectManager.py, to make Folders render nicely. # XXX hack to make folders render # retains compatibility with current behaviour # if the folder doesn't contain index_html def __call__(self, *args, **kw): """ mandatory docstring """ index = self._getOb('index_html', default=None) if index is not None: args = (self,) + tuple(args[1:]) return apply(index, args, kw) return '' Is this a good idea? Should this go into the Collector as an RFE with patch? Actually, I'm not sure why it should be tuple(args[1:]) -- I just copied that out of ZCallable in the latest PTK CVS. I'd guess you'd want to check the first argument (client) for whether it is None before replacing it with self. Perhaps something like this: args = (len(args)>0 and args[0] or self,) + tuple(args[1:]) Discussion welcomed. -- Steve Alexander Software Engineer Cat-Box limited http://www.cat-box.net

1 0

Announcing dev.zope.org
by Brian Lloyd 14 Jul '00

14 Jul '00

Hello all, For some time now, it has been a goal of ours to open up the Zope development process. Some of you may have heard me promise this at the Zope track back in January. Well, it took a little longer than I had hoped it would, but I'm happy to announce an important step in making that a reality: http://dev.zope.org. dev.zope.org is the place for discovering, initiating, and contributing to work on the core Zope platform and related Zope technologies. You can think of it as the rough equivalent of what Mozilla.org is to Mozilla. Why has it taken so long? Because there is much more to "opening the development process" than simply doling out CVS access. Evolving a relatively large and complex piece of software like Zope is a non-trivial task, and doing it in a highly distributed environment is harder still. The dev.zope.org site will provide the background materials and tools to ensure that work on Zope is consistent in organization and execution. It will also help those new to Zope development to come up to speed and get involved quickly. There is still plenty to do be done on the site, but it is done enough to start using it so I want to start learning by doing. If you have questions or comments on the dev site, please send them to me at brian(a)digicool.com and CC the zope-dev list (which is where I expect traffic related to dev.zope.org should go in the near term). Brian Lloyd brian(a)digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com

1 0

Namespace Questions
by Adam Pawliuk 14 Jul '00

14 Jul '00

Hi, I want to find an efficient way of adding a dictionary of values to the Zope namespace, where a different dictionary could be set on each request. I wanted to create a several dictionaries each containing properties for a look and feel (i.e. background colors, fonts etc), and switch them for each request. For example I'd like to do something similar to what's done below (although the method below doesn't work). **************************************************************************************** <dtml-call "REQUEST.set('LookAndFeel',m_getRandomLookAndFeel())"> (assuming the Look and Feel Dictionary has key's LOOK_COLOR, and LOOK_FACE) <dtml-with LookAndFeel> <font color="<dtml-var LOOK_COLOR>" face="<dtml-var LOOK_FACE>"> I know the with doesn't work for the dictionary, but would like to do something similar to this example. </font> </dtml-with> **************************************************************************************** thanks in advance, Adam

1 0