On Wednesday, October 17, 2001, at 05:11 PM, templates-
[EMAIL PROTECTED] wrote:

> The problem is, no two people on earth agree about what a "content
> management" system is.  Some people think it's the ability to edit web 
> pages
> through a browser text field.  Others think its version control.  Still
> others think it's workflow and templating for data.  As you've 
> discovered
> though, about 99% of the open source world thinks 'Content Management
> System' eq 'Slashdot'.
>
> Building a useful system would be fun.  I've considered building 
> something
> to handle static files by hooking up mod_perl with CVS and providing a 
> web
> GUI for basic site workflow.  I've also considered building a system 
> that
> provides workflow, scheduling, and versioning for user-defined data 
> objects,
> probably using OpenInteract.  The trouble is, with so little agreement 
> about
> goals and requirements, it's hard to get started without an actual 
> project
> that needs it.

Does anyone out there remember the project I was working on?  Iaido (nee 
Iaijutsu)?  Unfortunately, I had to mothball it so I could work on 
projects at a new job and have a life at the same time.  But it looks 
like I'm about to leave that job now and go to a new one where I might 
actually have some free time to work on it.  (The saddest thing is that 
the Next Big Thing I've been working on gets to die where I leave it :( )

Mostly mentioning it to see a) if anyone remembered it, b) if anyone 
would be interested in seeing it live again, and c) if anyone's 
interested in nailing down some goals and requirements, as Perrin says.  
Reason I mention it here is because TT was one of the primary reasons it 
was actually working well.

Anyway, short description:  Take an object persistence framework that 
works both as LDAP-like attribute directory and virtual filesystem.  
Take a web app server, with sessioning and other goodies, that maps 
request URLs onto object lookups and method invocations on those 
objects.  Persistent objects have methods called by URL requests, which 
display results using TT2.  Templates used by the objects have a 
Javascript-like interface to the rest of the framework.  (ie. a list of 
latest news items displayed by a news folder object might use something 
like [% FOREACH item = self.contents %] to gain contents to itself, and 
might have [% parent.url %] to construct a link to the parent section of 
the site).  URL-accessible Methods on all objects are mediated by an ACL 
system, and content is web-editable, FTP accessible, and I even had an 
experimental Perl FS mounting under Linux.

It was trying to compete with Zope in Python.  I was getting close on 
versioning, workflow, and scheduling, and had a lot of other neat things 
lined up.  It also has a lot of very braindead stuff in it, so I was 
thinking of starting over with the same concept because I still think a 
lot of the ideas were solid.  I keep thinking I might want to start over 
in Java, but I keep forgetting why I think that...

Anyway, any interest?

--
Leslie Michael Orchard <[EMAIL PROTECTED]>
ICQ: 492905 (home) 11082089 (work)
"...see you space cowboy..."



Reply via email to