August 2000 - Zope - Zope lists

My notes from the Paris EuroZope meeting
by Simon Coles 30 Aug '00

30 Aug '00

Hi, During the EuroZope meeting I was taking notes for my own use, and a couple of people asked if I would make them more widely available - so here they are. Because they were taken during the session (and I've got RSI :-) they are bitty and may not make sense. They are also only a reflection of what I typed, not of what was said and its importance :-) so any mistakes are my fault and I humbly apologise now. Mostly these notes cover Paul's talk in the morning, as my machine was the IRC terminal for the afternoon sessions. Anyway, for what its worth, here you are.... ---------------------------------------------------------- The meeting: - 60 People turned up - overflowed the room and it became very warm :-( - 50% of the people there use Linux as their primary OS. Paul's talk =========== (PE = Paul Everitt, OSS = Open Source, DC = Digital Creations) (show of hands) Most people view the fact that Zope is free and Open Source is the major thing about it. Believes DC has a sustainable business model for OSS - Red Hat etc. don't. Lots of stuff in Zope originally comes from DC's work with Newspapers - users being able to edit stuff, undo, etc. DC and business model - DC were spun out of a newspaper consortium who considered the object stuff "worthless" so gave it to DC when they were spun out. - They realised they needed to be a service/consulting company - not a product company - because the market was already done on web application servers and it would require *massive* amounts of investment to even have a chance of competing. - Their first investment was a banker who used the stuff and then called them to invest. Following that minority investment he convinced them to release it all as OSS. The rest is history etc. - First round investment, spent it on Zope, not DC. DC != Zope. Because then Zope doesn't appear commercial, protects the OSS nature. They don't want to dominate Zope, but they want to back it. - People are throwing money at DC, seems like intelligent money not any old VC trying to get on bandwaggon. 2nd round of money spent on Zope, 3rd round of money spent on DC. - In '99 made no money. In '00 they turned sales on, and all hell broke loose - Zope is hot property, "people were beating themselves up in the sales funnel". - DC is 30 people. - Additional investments in DC are on the horizon, which should mean more of everything for Zope From Business Partner's perspective will get more formalised etc. Brian Lloyd was employee number 1 of DC, now product manager of Zope. His number 1 goal is to open up Zope - make Zope bigger than DC. Giving other people checkin authority is a big aim. They do OSS distribution but not OSS development at the moment (because DC do the development). Therefore number 1 priority. DC are involved in Mozilla and are getting a good idea of how they run OSS. He later demo'd the Zope studio, and it looks interesting. Zope will still have a benevolent dictator model. They come out of the Python community so will be like that. Was some discussion around how to run large OSS projects etc. Big things will happen in the next 6 months. Lots of opportunity, sky's the limit etc. What do you want in Zope 3? "The Soul of Zope" - how do you describe Zope, where its going etc. Very hard. ........"Focussed on rapid deployment of rich solutions to diverse audiences". This drives into 3 things... (see later - "More power for more people", "All tiers as managed content", "Web API") Objects on the web that ordinary people can walk up to and do rich and interesting stuff - and then they can talk those objects and do their own thing. Diverse - so everyone in the food chain. Things that Work the way people work. Allow people to recover from their mistakes. More power for more people - safely delegating control, use web as complete object system. Security, undo, versions, transactions - all those things that stop people from screwing things up. All tiers as managed content. Similar to 3 tier model. The way all the existing app servers do is wrong, the way they are doing it right. Presentation and logic on file system - hierarchical storage model, messy to admin etc. When you want to cluster, you need to cluster the file system - hard/expensive. presentation managed differently from logic (different tools). Different data model for data ("square" - tabular, not hierarchical). So each layer completely different and independent. Zope wants to treat each layer in the same way. Will all look and feel very similar. Cluster, manage, search etc. Presentation, logic, data. This is a sweet spot for Zope. Gives a scorecard for Zope design decisions. Tools for developing objects GoLive 5 have great WevDAV. WEbDAv will get a lot better. Python products could be endangered species because its managed in the file system, which isn't in the "all tiers as managed content" thing. Web API SOAP is big as well. Tim O'Rielly - "should Amazon have an API?". Zope is better at the API thing than everything else out there, as the entire API is exposed to things like XML-RPC. Microsoft's .NET is going to happen, OSS will have an alternative, Zope wants to participate in that. Zope could become the OSS .NET server alternative. Zope futures ------------ Major thing is usability which involves: - Documentation - User Interface - Concepts - try to reuse concepts elsewhere rather than coming up with their own Next thing is hardening - security, reliability Have Oracle ZODB back end now, they are deploying it now, will release but want to document On concepts - there's enough in XML DOM to mean they don't have to live in their own little world. Maybe should use the DOM instead of the Zope API. Also schemas, component models. Therefore want to turn Zope from an app server into an XML application server, so if you know XML you will feel very at home in Zope. Therefore they don't have to write docs, they can just send people to W3C :-) Will use 4Thought's DOM (standard DOM for Python), ZDOM will go away. XML will be the interface, but internal to Zope for performance reasons, it will be normal Zope objects. Your entire Zope site will look like one big XML document. XML object database with the trimmings (cache, transactions, distributed etc.) will be very compelling for XML people. They hope that XML will be considered safer than Oracle (Vs ACS, which uses Oracle) Other major effort is PTK - but the name won't last much longer. Meant to be beginning of strategy for DC & Zope where Zope is a platform with many products. But been confused - "Should Zope be a product" - and useful out of the box. But for who? Hence decide that Zope will be a facility for creating solutions, and won't be the product. Therefore UI and docs for Zope will be focussed on a particular audience (developers). Problem in the past was they couldn't say no. PTK will be a product running on Zope, and will allow DC to get into the Content Management market, where they will make their money. CM market is dominated by StoryServer - very large prehistoric beasts, proprietary lock-in model. Pay $1M to get something that doesn't do what you want to do, then another $1M to customise. Feel OSS will go over very well in the CM market rather then DB or others. Because CM is real bet the business decision, hence OSS because more control. CM companies make their money on their services, therefore OSS the product won't lose revenue. Expect interesting/wierd things to happen with PTK, ZEO, and ASP model. With PTK, Create something that's useful out of the box, other people working on it beside DC, focus on site designers not developers so much. Keen to understand what people want from Zope 3, what they would do if they were PE. PTK is a product on Zope platform which is immediately useful for the problem domain of content management and portals. Shared set of services and then people will work with those things either by being a member or a manager of a site. PTK --- (quite a lot of PTK stuff was dealt with in the above talk) all default content is catalog aware in the ptk, the search engines looks different depending on who you are - depending on security roles etc. (bit of business partner stuff thrown in to PTK talk) They have referred out 15 bits of business. Feel that things like PTK need a large developer/consulting community. PTK architecture now being used be DC consulting contracts. Wizards look cool. PTK UI is worth looking at. Calling it a Portal was a mis-name. Was called Portal before Portals became something. Expect new name. Are they planning to move into ECommerce like Vignette? "It has to be there to be competitive, but DC need help on it. Does someone want to write a spec for it?" Groups != Roles. Roles are really what is important, hence although its harder for people to wrap their head around, they will stay with Roles. But got some objections and good reasons why there should be groups - listened attentively - therefore asked please write up for dev.zope.org They are taking requests for Zope 2.3 now, so get writing. Business partner programme -------------------------- They have had lots of interest Creating one is very hard Its very important to them that Zope > DC They want to match customers to businesses in a way that's good for everyone 3/4 of people at the meeting were interested in the business partner programme DC are currently occupied by 4 major customers. What do DC want from Business Partner programme: - "Better revenue" - allow DC to focus on platform work, as a tech partner/provider - Different levels of membership (with different levels of attention from DC) - Quickly grow the number of solutions on the platform Continue to focus Zope on business issues, rather than on technology - in contrast to most OSS projects which are generally about technology with little business relevance. What do businesses want from Business Partner programme: - Leads - Certification There is a private mailing list for the Business Partner programme. Certification will have implications for the Zope release schedule (as will other things, like increasing popularity). They are going to move to something more like the Linux Kernel release scheme, with development and stable versions. Business Partner programme is very important revenue opportunity for DC, but not decided how yet. Support programmes are the main thing. ---------------------------------------------------------- Overall, it was an excellent meeting and well worth attending. Thanks to all of those who made it happen, to Republic Alley (http://www.republicalley.com) for the use of their offices, and Paul for spending his holiday with us (and Paul's wife for letting him :-). Simon -- --------- My opinions are my own, NIP's opinions are theirs ---------- Simon J. Coles Email: simon(a)nipltd.com New Information Paradigms Work Phone: +44 1344 753703 http://www.nipltd.com/ Work Fax: +44 1344 753742 =============== Life is too precious to take seriously ===============

2 1

Re: [Zope] ZCatalog in 2.2.x
by Dieter Maurer 30 Aug '00

30 Aug '00

Terry Kerr writes: > .... But now when I instantiate my product in 2.2.0 or 2.2.1, > searching only works when I search using a field index. Searching using > a textindex always results in an empty result .... I use ZCatalog text indexes under 2.2.1b1 without problems (with a globbing lexicon!). I had to fix some glitches (search for "ZCatalog glitches" and "Globbing Lexicon bug" in the Collector). However, none of my fixes had to do with your problem description. Dieter

1 0

Re: [Zope] symbolic links or invisible objects..?
by Dieter Maurer 30 Aug '00

30 Aug '00

Geoffrey L. Wright writes: ... symbolic links ... If your content is simple, i.e. does not contain URL references, then symbolic links can easily be emulated with DTML methods. Suppose, you have the following structure: A/ A/A1/ A/A1/a1 .... A/A1/an A/B1/b1 .... A/B1/bn and you want "A/B1/bi" be a symbolic link to "A/A1/ai", then "bi" would be the following DTML method: <dtml-with "A.A1"><dtml-var ai></dtml-with>. This will not work, if "ai" make a URL reference to an "aj" unless "aj" and "bj" have the same id and "bj" is a symbolic reference to "ai". For more complex situations, you may use the product SiteAccess (--> zope.org) to redirect requests transparently according to situation. You may look at URL:http://www.dieter.handshake.de/pyprojects/zope/Localizer.py This is used as a SiteAccess AccessRule for localization. Requests going to a folder controlled by this rule are served from this folder, if possible. If not possible, the request is redirected to a different folder. I think, this method can be abused to get what you want. Dieter

1 0

Re: [Zope] FTP crashing Zope?
by Dieter Maurer 30 Aug '00

30 Aug '00

Dario =?ISO-8859-1?Q?Lopez-K=E4sten?= writes: > Using FTP Explorer I crashed (killed) my Zope instance while ftping in to > port 8021. This was achieved by not setting FTP Explorer to use PASSIVe > connections. Using Passive it works like a charm. > > Is this a bug in latest Zope (2.2.1) or is this expected behaviour? If, the > latter is the case, what can I do to a) turn off FTP access, b) make sure > that my Zope is up and running automagically after a crash? a) look at the documentation in "z2.py". It tells you, how to selectively enable/disable protocols. b) When my Zope had crashed, it usually restarted automatically. Dieter

1 0

Re: [Zope] ZClass <--> Product
by Karl Anderson 30 Aug '00

30 Aug '00

Chris Withers <chrisw(a)nipltd.com> writes: > Karl Anderson wrote: > > What FTP editing options are there for ZClass definitions? > > Your email address suggests you should be able to find out more easily > than me ;-) Well, I'm pretty sure the answer is "none", but I'm asking to the list so if someone does have some kind of workaround, the reply will be public. > Maybe you could suggest better editing support for FTP or is WebDAV the > way of the future? > Mention was made of Adobe GoLive! in Paris, what's that like for doing > this sort of this? FTP is limited for this sort of thing because it works on files with no metadata. I *think* that the only reason one can't edit DTML methods in a ZClass definition is that the intermediate steps don't support a FTP directory interface. That'd be a nice first step. I've heard some ideas about how we could edit properties, permissions, etc. with FTP, some more hackish than others. Best is probably some sort of XML text file associated with an item. But from what I've been reading, the focus will be on through the web editing whenever possible. I personally find it much more convenient to work with python baseclasses & have a python shell looking at the ZODB in an emacs buffer, so I have a history of what I've done and seen and can look at the objects & attributes on any level. And yes, the wave of the future does appear to be with the various clients that are coming out, Radio Userland being the most recent that I've heard about. -- Karl Anderson karl(a)digicool.com

1 0

Export/import problem
by Oras Kaarel (2914) 30 Aug '00

30 Aug '00

Hi I have two different Zope types (Zope 1.10.3 (binary release, python 1.5.1, win32-x86) and Zope version: Zope 2.2.1 (binary release, python 1.5.2, win32-x86)) and our intranet is located in the older version. I thought that exporting/importing would be a great idea. I managed to export *.bbe file and I copied it in import catalog in new Zope, but the new version of Zope announced, that: ---------------------------------------------------------------------------- Zope has encountered an error while publishing this resource. Error Type: ExportError Error Value: Invalid export header Troubleshooting Suggestions * The URL may be incorrect. * The parameters passed to this resource may be incorrect. * A resource that this resource relies on may be encountering an error. For more detailed information about the error, please refer to the HTML source for this page. If the error persists please contact the site maintainer. Thank you for your patience. ---------------------------------------------------------------------------- ------------------- Traceback (innermost last): File C:\PROGRA~1\WebSite2\lib\python\ZPublisher\Publish.py, line 222, in publish_module File C:\PROGRA~1\WebSite2\lib\python\ZPublisher\Publish.py, line 187, in publish File C:\PROGRA~1\WebSite2\lib\python\Zope\__init__.py, line 221, in zpublisher_exception_hook (Object: ApplicationDefaultPermissions) File C:\PROGRA~1\WebSite2\lib\python\ZPublisher\Publish.py, line 171, in publish File C:\PROGRA~1\WebSite2\lib\python\ZPublisher\mapply.py, line 160, in mapply (Object: manage_importObject) File C:\PROGRA~1\WebSite2\lib\python\ZPublisher\Publish.py, line 112, in call_object (Object: manage_importObject) File C:\PROGRA~1\WebSite2\lib\python\OFS\ObjectManager.py, line 504, in manage_importObject (Object: ApplicationDefaultPermissions) File C:\PROGRA~1\WebSite2\lib\python\ZODB\ExportImport.py, line 143, in importFile ExportError: (see above) Can anyone pls help me in this case pls --------------------------------------------------------------------- Kaarel Oras

2 1

Overriding server's notion of SERVER_NAME, SERVER_PORT, etc.
by Skip Montanaro 30 Aug '00

30 Aug '00

In 2.2.1 I can no longer override ZServer's notion of SERVER_NAME, SERVER_PORT, etc. from the invoking shell environment. (I am currently running 2.0.0, so this may well be true of other earlier 2.1.x and 2.2.x versions. With a little luck, I may actually be able to upgrade in the next few days...) I run ZServer listening to 127.0.0.2, port 9673. The only thing that talks directly to it is my front-end Apache server via its proxy module, so I want any HTML generated to reflect the visible address and port, not ZServer's hidden address and port. I was previously able to force these variables to be what I wanted by simply setting them before running the server: SERVER_NAME=dolphin.mojam.com ; export SERVER_NAME SERVER_PORT=80 ; export SERVER_PORT HTTP_HOST=$SERVER_NAME:$SERVER_PORT ; export HTTP_HOST SCRIPT_NAME=/ ; export SCRIPT_NAME but 2.2.1 ignores these settings. I made a few simple changes to ZServer/HTTPServer.py and lib/python/ZPublisher/HTTPRequest.py that use environment variables as defaults if they're available. Context diffs are attached to this msg. I'd appreciate feedback from those in the know if these were the correct sort of changes to make or if there is a better (supported) way to achieve my goal. Thx, -- Skip Montanaro (skip(a)mojam.com) http://www.mojam.com/ http://www.musi-cal.com/ diff -rc Zope-2.2.1-src/ZServer/HTTPServer.py Zope-2.2.1-src-new/ZServer/HTTPServer.py *** Zope-2.2.1-src/ZServer/HTTPServer.py Tue Jun 27 10:55:14 2000 --- Zope-2.2.1-src-new/ZServer/HTTPServer.py Wed Aug 30 13:47:12 2000 *************** *** 223,238 **** server=request.channel.server env = {} env['REQUEST_METHOD']=upper(request.command) ! env['SERVER_PORT']=str(server.port) ! env['SERVER_NAME']=server.server_name env['SERVER_SOFTWARE']=server.SERVER_IDENT env['SERVER_PROTOCOL']=request.version env['channel.creation_time']=request.channel.creation_time if self.uri_base=='/': ! env['SCRIPT_NAME']='' ! env['PATH_INFO']='/' + path else: ! env['SCRIPT_NAME'] = self.uri_base try: path_info=split(path,self.uri_base[1:],1)[1] except: --- 223,242 ---- server=request.channel.server env = {} env['REQUEST_METHOD']=upper(request.command) ! env['SERVER_PORT']=os.environ.get("SERVER_PORT", str(server.port)) ! env['SERVER_NAME']=os.environ.get("SERVER_NAME", server.server_name) env['SERVER_SOFTWARE']=server.SERVER_IDENT env['SERVER_PROTOCOL']=request.version env['channel.creation_time']=request.channel.creation_time if self.uri_base=='/': ! env['SCRIPT_NAME']=os.environ.get("SCRIPT_NAME", '') ! if not env['SCRIPT_NAME']: ! env['PATH_INFO']='/' + path ! else: ! env['PATH_INFO'] = env['SCRIPT_NAME'] + path else: ! env['SCRIPT_NAME'] = os.environ.get("SCRIPT_NAME", ! self.uri_base) try: path_info=split(path,self.uri_base[1:],1)[1] except: diff -rc Zope-2.2.1-src/lib/python/ZPublisher/HTTPRequest.py Zope-2.2.1-src-new/lib/python/ZPublisher/HTTPRequest.py *** Zope-2.2.1-src/lib/python/ZPublisher/HTTPRequest.py Tue Aug 15 15:07:37 2000 --- Zope-2.2.1-src-new/lib/python/ZPublisher/HTTPRequest.py Wed Aug 30 13:43:05 2000 *************** *** 266,272 **** else: b='' while b and b[0]=='/': b=b[1:] ! server_url=get_env('SERVER_URL',None) if server_url is not None: other['SERVER_URL'] = server_url = strip(server_url) else: --- 266,272 ---- else: b='' while b and b[0]=='/': b=b[1:] ! server_url=get_env('SERVER_URL',os.environ.get("SERVER_URL", "")) if server_url is not None: other['SERVER_URL'] = server_url = strip(server_url) else:

1 0

ZClass and Propertysheets
by Andy Pahne 30 Aug '00

30 Aug '00

Hi, I have created a ZClass with a propertysheet called "Basic". I added a property "db_id", integer value with a default value of 0. Then I created two instances and changed the value to 1 in the first and 2 in the second instance. When I try to render the property with a <dtml-var db_id> I get a 1 in both instances. That's not what I expected in the beginning. In particular not because there doesn't seen to be anything wrong when I have a look at the property sheet in the instance's management screen. I have different values there. Am I wrong or is there something wrong with Zope 2.2.1 and ZClasses? Andy Pahne ap(a)big1.de

1 0

RE[3]: [Zope] Product inheritance question. ==> Pb. moving clas s from a product to another + etc. s from a product to another + etc.
by NABETH Thierry 30 Aug '00

30 Aug '00

Hello, >> If it is a class method, it must be inherited by any derived class. This is what I had done. However, I had the time finally to conduce the following tests. - create a Product P1 (Note: I used the word "package" instead of Product in my previous postings) - create a Class A in P1 - create a method z in the Class A - create a Product P2 - create a Class AA that inherits A in product P2. - create an object 'o' instance of Class AA. - make the test of calling the method z on 'o'. Results: everything works fine !! So everything seems to work correctly with the new version. However, when I try to do the same (create a method z and test) for Packages that I had created in the previous version of Zope, it just does not work !! (I have imported in the new version the class I had created in the previous version). I believe that this had to do with the trick I used to split and existing Product in several Products: Since I could not move a class directly from a package to another, I created 2 copies of the same packages. I deleted from the first package the class I wanted to move. From the second package, I deleted the classes I didn't want to move. The result was 2 packages with each the correct set of classes. The problem is that this seems to have generated some kind of inconsistancy, in particular related to the namespace: the second package do not have access to the namespace of the first package. I know this was a relatively ugly way to proceed, but I found at that having a zillion of classes in a same package was even more ugly, and I needed to reduce this complexity by introducing more modularity. Of course, at the beginning I couldn't predict which Products and which classes would exist and I decided to start with one package and split it later. If you had nor noticed I used some kind of incremental/prototypical/evolutionary/blablabla methodology. Unfortunatly, not being able -- to move classes from one product to another -- makes the use of this methodology problematic :-(. Another similar problem is that you can not modify a posteriori the inheritance tree. (for instance when you decompose a class into 2 subclasses that inherit from one another). To conclude, I find Zope to be a very interesting systems (because it is dynamic, object-oriented, integrate seamlessly an OO database, etc.). However, it would be an even better system if it provided more flexibility to reengineer an architecture. (very important when you use a prototypical approach). Or perhaps, there is something I have not understood ? Thierry Nabeth Research Fellow INSEAD CALT (the Centre for Advanced Learning Technologies) http://www.insead.fr/CALT/ -----Original Message----- From: Dieter Maurer [mailto:dieter@handshake.de] Sent: Monday, August 21, 2000 10:47 PM To: thierry.nabeth(a)insead.fr Subject: Re: RE[2]: [Zope] Product inheritance question (similar question) NABETH Thierry writes: > What I will have to do: > - create a Package (Zope product). > - create a class in this package, and a method (my_method) This should be a method of the class, not a method in the packages. If it is a class method, it must be inherited by any derived class. Dieter -----Original Message----- From: NABETH Thierry Sent: Friday, August 18, 2000 9:29 AM To: 'zope(a)zope.org' Subject: RE: [Zope] Product inhetitance question (similar question) And what happen if the Class A is in a package PA. Clabb B is in a package PB. How do you access the namespace of PA from PB ? When I have tried, the inherited methods from A where not visible from B, and I and to recreate a copie of the method. (which is ugly !!!). Any suggestion ? Thierry Nabeth Research Fellow INSEAD CALT (the Centre for Advanced Learning Technologies) http://www.insead.fr/CALT/

1 0

Generic Workflow Products?
by Cary O'Brien 30 Aug '00

30 Aug '00

After applying Zope to a couple of in-house projects (which turned out pretty well, thank you), I've started to come to the realization that many applications boil down to workflow management. A job comes in, attributes get set, things get attached to it, and it gets passed from person to person (to machine, in some cases) as it flows through the system. I guess to start out it would need the following: -- user classes -- job creation -- job attributes, with protection from view and modification depending on user classes -- multiple job queues, with access control, and some sort of priority. Queues may end up being implemented simply by having a job state, sequence number, and priority, and moving from one queue to another would just be changing the state. -- job timeouts for escalation, -- e-mail "ticklers" and notification -- job routing based on attributes and actions. -- some sort of api other then HTTP/HTML for external programs to access jobs, perform actions, and update jobs. -- Extra Credit: Graphical workflow editor. -- (what else am I missing) Sounds like a job for Zope, doesn't it. Basically I've implemented a couple of systems like this, either classic database-driven or z-class driven. But I'm tired of re-inventing the wheel. It would be nicer if systems like this could be implemented at the "workflow logic" level rather than the Zclass level. IE a set of ZClasses or Product classes that implemented workflow systems. Is there such a thing? Has anyone worked on such a thing? Any ideas how to implement such a thing? I've got a couple of modifications to an existing system coming up, which sounds like just the excuse I need for a total rewrite :) -- cary

5 9