Re: [Zope-dev] Branches finished for merging.
Do we want to include optional twisted in 2.10 too? It works now, for http. :-) Haven't tried the ftp yet, but if I finish it until tomorrow? ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
On 4/26/06, Chris Withers <[EMAIL PROTECTED]> wrote: > Lennart Regebro wrote: > > I also recommend that we deprecate both __bobo_traverse__ and > > __browser_default__, but perhaps with a longer derecation period that > > usual, since this are very basic techniques used in many products. > > Please discuss. :) > > -1 - what's wrong with having the default adapter around "forever" for > Zope 2? It's ugly? :) Or rather, having the default adapter around forever is a definite yes. Because it does attribute and key look up as well. But the code could be simplified. HOWEVER, I'm not gonna have time to remove all uses of __browser_default__ and __bobo_traverse__, so we don't have to discuss it more closely for months, because it won't be deprecated in 2.10 in any case. ;) -- Lennart Regebro, Nuxeo http://www.nuxeo.com/ CPS Content Management http://www.cps-project.org/ ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
--On 27. April 2006 10:33:19 +0200 Lennart Regebro <[EMAIL PROTECTED]> wrote: On 4/27/06, Andreas Jung <[EMAIL PROTECTED]> wrote: Getting rid of ZServer is a fine goal but as long as Twisted is much slower (from what I've heard) ZServer has to stay (ZServer is not subject to be deprecated at this time). ZServer will remain the default for now, and it will not be deprecated unless the community wishes so. I'm guessing the community will want to sooner or later. ;-) Sure, we all want that (but not for the sake of slower performance) :-) Twisted is slower, but how much of a difference that makes in real life applications I don't know. The server is in itself not a very big performance eater in Zope. There is of course also the option to use something else that can use WSGI, like for example zope.server, which is much cleaner than ZServer, and probably therefore easier to maintain. Can I only repeat the "rumors" I've heard from the integration of Twisted in Zope 3. So we won't give up an "efficient" implementation of a core component..that's just what I wanted to say. Having some benchmarks would be nice. And btw. thanks for your work :-) Andreas -- ZOPYX Ltd. & Co. KG - Charlottenstr. 37/1 - 72070 Tübingen - Germany Web: www.zopyx.com - Email: [EMAIL PROTECTED] - Phone +49 - 7071 - 793376 E-Publishing, Python, Zope & Plone development, Consulting pgpYJSXJUkIIm.pgp Description: PGP signature ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
On 4/27/06, Andreas Jung <[EMAIL PROTECTED]> wrote: > Getting rid of ZServer is a fine goal but as long as Twisted is much slower > (from what I've heard) ZServer has to stay (ZServer is not subject to be > deprecated at this time). ZServer will remain the default for now, and it will not be deprecated unless the community wishes so. I'm guessing the community will want to sooner or later. ;-) Twisted is slower, but how much of a difference that makes in real life applications I don't know. The server is in itself not a very big performance eater in Zope. There is of course also the option to use something else that can use WSGI, like for example zope.server, which is much cleaner than ZServer, and probably therefore easier to maintain. > > The risks is that somehow we introduce some subtle behavioural change > > that isn't tested for. > > I would love to see the stuff tested with the major frameworks (CMF 2, > Plone + CPS) before it will be merged into the trunk. Sure. > > The traversal refactoring means that the role checking is now done > > after each traversal step, instead of intermingled with it. The > > functionality should be the same, but perhaps there is some untested > > special case when it changes. We have had much larger incompatibility > > changes in earlier major releases though. > > > > The WSGI refactoring may mean that somebody using internal features of > > ZServer might need to change something. Also it means that what the > > publish_module() method expect as parameters will change. > > > > I have no problems with the traversal refactoring..that sound sane. -- Lennart Regebro, Nuxeo http://www.nuxeo.com/ CPS Content Management http://www.cps-project.org/ ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
--On 27. April 2006 00:17:18 +0200 Lennart Regebro <[EMAIL PROTECTED]> wrote: On 4/26/06, Andreas Jung <[EMAIL PROTECTED]> wrote: What are the possible risks of those changes? What is the goal of this branch? Integration as an optional feature or replacement of ZPublisher/ZServer? The direct goal of this is to make it more pluggable, so that we can introduce twisted or zope.publisher as alternatives. Twisted, so that we can write new servers for Zope2, and zope.publisher, so that we can closer to the goal of being able to run both Zope3 and Zope2 software under the same server. That's fine The long term goal is to replace ZServer/ZPublisher with whatever Zope3 uses, which at that point in time most likely will be twisted/zope.publisher, to tighten the Z2/Z3 integration, and reduce the amount of code we need to maintain. Getting rid of ZServer is a fine goal but as long as Twisted is much slower (from what I've heard) ZServer has to stay (ZServer is not subject to be deprecated at this time). The risks is that somehow we introduce some subtle behavioural change that isn't tested for. I would love to see the stuff tested with the major frameworks (CMF 2, Plone + CPS) before it will be merged into the trunk. The traversal refactoring means that the role checking is now done after each traversal step, instead of intermingled with it. The functionality should be the same, but perhaps there is some untested special case when it changes. We have had much larger incompatibility changes in earlier major releases though. The WSGI refactoring may mean that somebody using internal features of ZServer might need to change something. Also it means that what the publish_module() method expect as parameters will change. I have no problems with the traversal refactoring..that sound sane. -aj -- ZOPYX Ltd. & Co. KG - Charlottenstr. 37/1 - 72070 Tübingen - Germany Web: www.zopyx.com - Email: [EMAIL PROTECTED] - Phone +49 - 7071 - 793376 E-Publishing, Python, Zope & Plone development, Consulting pgpReGz3xMqf3.pgp Description: PGP signature ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
On 4/26/06, Andreas Jung <[EMAIL PROTECTED]> wrote: > What are the possible risks of those changes? What is the goal of this > branch? Integration as an optional feature or replacement of > ZPublisher/ZServer? The direct goal of this is to make it more pluggable, so that we can introduce twisted or zope.publisher as alternatives. Twisted, so that we can write new servers for Zope2, and zope.publisher, so that we can closer to the goal of being able to run both Zope3 and Zope2 software under the same server. The long term goal is to replace ZServer/ZPublisher with whatever Zope3 uses, which at that point in time most likely will be twisted/zope.publisher, to tighten the Z2/Z3 integration, and reduce the amount of code we need to maintain. The risks is that somehow we introduce some subtle behavioural change that isn't tested for. The traversal refactoring means that the role checking is now done after each traversal step, instead of intermingled with it. The functionality should be the same, but perhaps there is some untested special case when it changes. We have had much larger incompatibility changes in earlier major releases though. The WSGI refactoring may mean that somebody using internal features of ZServer might need to change something. Also it means that what the publish_module() method expect as parameters will change. -- Lennart Regebro, Nuxeo http://www.nuxeo.com/ CPS Content Management http://www.cps-project.org/ ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
Lennart Regebro wrote: I also recommend that we deprecate both __bobo_traverse__ and __browser_default__, but perhaps with a longer derecation period that usual, since this are very basic techniques used in many products. Please discuss. :) -1 - what's wrong with having the default adapter around "forever" for Zope 2? cheers, Chris -- Simplistix - Content Management, Zope & Python Consulting - http://www.simplistix.co.uk ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
--On 26. April 2006 19:14:43 +0200 Lennart Regebro <[EMAIL PROTECTED]> wrote: 1. easter-sprint_traversal-refactor I also recommend that we deprecate both __bobo_traverse__ and __browser_default__, but perhaps with a longer derecation period that usual, since this are very basic techniques used in many products. Please discuss. :) +1, also on a longer deprecation period. 2. regebro-wsgi_refactor The work Michael and I did last week also included an effort to finish Sidnei da Silvas publication traversal branch, that attempts to replace ZPublisher with zope.publisher and ZServer with twisted. Trying to finish that work, I grew restless with the tight integration between all the modules, and have therefore tries to pry ZPublisher and ZServer apart, but wedging WSGI in between them. The modules are not 100% separated as the ZPublisher still imports some modules from the ZServer and the other way around, but there is now a WSGI application class that should be callable from any WSGI server. I propose to check this in to make it possible to use twisted with Zope2, even though we may not have time to include the twisted server in the release. What are the possible risks of those changes? What is the goal of this branch? Integration as an optional feature or replacement of ZPublisher/ZServer? Andreas pgpPTbZIkg887.pgp Description: PGP signature ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
Re: [Zope-dev] Branches finished for merging.
Em Qua, 2006-04-26 às 19:14 +0200, Lennart Regebro escreveu: > I have (at least) two branches that will be merged before The Big > Freeze on monday, if all goes well, and unless somebody has any very > serious objections. > > 1. easter-sprint_traversal-refactor > [...] > 2. regebro-wsgi_refactor +1 to merging both branches > I also recommend that we deprecate both __bobo_traverse__ and > __browser_default__, but perhaps with a longer derecation period that > usual, since this are very basic techniques used in many products. > Please discuss. :) +1 to deprecation and to longer period for this case. -- Leonardo Rochael Almeida Enfold Systems http://www.enfoldsystems.com phone. +1.713.942.2377 Ext 215 fax. +1.832.201.8856 ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )
[Zope-dev] Branches finished for merging.
I have (at least) two branches that will be merged before The Big Freeze on monday, if all goes well, and unless somebody has any very serious objections. 1. easter-sprint_traversal-refactor This was done by me an Micheal Kerrin during last weeks Sprint here in Paris. It involves a backwards compatible refactoring of the Zope2 publishing traversal. It will for special traversing of names and for default publishing traversal now use an IBrowserPublisher adapter. Objects wishing to override the default traversal would earlier have has a special __bobo_traverse__ method. They should now instead provide an IPublishTraversal adapter (or IBrowserPublisher, which extends IPublishTraversal). For backwards compatibility, any object that does not use an adapter will get a default adapter that tries to use __bobo_traverse__. Objects wishing to provide a special default name would easlier have a special __browser_default__ method. They should now intead provide an IPublishTraversal adapter. For backwards compatibility, any object that does not use an adapter will get a default adapter that tries to use __browser_default__. I also recommend that we deprecate both __bobo_traverse__ and __browser_default__, but perhaps with a longer derecation period that usual, since this are very basic techniques used in many products. Please discuss. :) 2. regebro-wsgi_refactor The work Michael and I did last week also included an effort to finish Sidnei da Silvas publication traversal branch, that attempts to replace ZPublisher with zope.publisher and ZServer with twisted. Trying to finish that work, I grew restless with the tight integration between all the modules, and have therefore tries to pry ZPublisher and ZServer apart, but wedging WSGI in between them. The modules are not 100% separated as the ZPublisher still imports some modules from the ZServer and the other way around, but there is now a WSGI application class that should be callable from any WSGI server. I propose to check this in to make it possible to use twisted with Zope2, even though we may not have time to include the twisted server in the release. Comments? -- Lennart Regebro, Nuxeo http://www.nuxeo.com/ CPS Content Management http://www.cps-project.org/ ___ Zope-Dev maillist - Zope-Dev@zope.org http://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope )