[Zope-PTK] Member Publishing

Kent Polk kent@goathill.org
14 Dec 2000 02:02:25 GMT


On 13 Dec 2000 09:30:00 -0600, Shane Hathaway wrote:
>Kent Polk wrote:
>> What type of object/transition tracking mechanism are you planning
>> for this?
>
>An attribute of portal content objects called workflow_status will hold
>workflow state.  The list of transitions is read from the state
>machines.

Great! Looking better all of the time. I'm hoping (assuming) the
transition list elements will be wrapped for the current user
permission/state to only allow acceptible options here. 

>> What type of mechanism would you use to trigger the state transition?
>> Rule-based or requirements-based or what? Would the current state
>> determine the transition or would a transition-manager?
>
>The user actions, like approving or editing a document, are the
>transitions.

Yeah!

I'm hoping further that:

1) transitions can also be selected programatically (i.e a user or
   portal might be a software component, including xmlrpc method
   support?)

2) the workflow objects can retrieve their associated data various
   ways (such as SQL query), using object attributes as keys,
   including maybe the object id? ...

We are laying out a trivialized model of a Publishing House Workflow
to see how this might work. In this model, each workflow creation
corresponds to a request, either for action or for information
(query). Each portal lists the outstanding workflow objects and
subobjects (std Zope tree layout), with some associated status
info.

As to :
< It would be very helpful if someone could create UI mockups of a state
< machine creation and editing interface.  We need to know whether state
< machines of this kind are simple enough to express on web pages.  If
< they are too complex, we can simply create an API and the state machine
< would be generated by Python calls to the API.

I'm not sure what kind of transition 'portal' is planned, but with
an adequate way to control which transition elements are available
according to the current portal view, it appears to us that it can
be done with web forms. Have to think this thing through a bit more
and need more info from you.

The Publishing House Model appears to be a pretty good model for
what you are trying to do as well as what we are trying to do,
given that the state machines can know how to retrieve the content
they are associated with in a general fashion (SQL, etc). Just that
the manuscripts we are trying to write are Projects (genotypic and
phenotypic representations of populations of individuals) instead
of books, our authors are programs, genotypers and statisticians,
and our editors are lab managers and P.I.'s.

Still on track here???