----- Original Message ----- From: "Peter Donald" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Cc: <[EMAIL PROTECTED]> Sent: Tuesday, April 17, 2001 11:39 PM Subject: Re: [VOTE] Task API
> At 10:42 17/4/01 +1000, Conor MacNeill wrote: > > > >----- Original Message ----- > >From: "Peter Donald" <[EMAIL PROTECTED]> > >To: <[EMAIL PROTECTED]> > >Cc: <[EMAIL PROTECTED]> > >Sent: Tuesday, April 17, 2001 9:09 PM > >Subject: Re: [VOTE] Task API > > > > > >> At 08:12 17/4/01 +1000, Conor MacNeill wrote: > >> >> * tasks should have access to its own XML representation. > >> >> > >> >> /Task Object Model instead of XML representation/ > >> > > >> >-1. > >> > >> How do you propose to do container tasks? > > > >What is a container task? (I could guess, just want to be clear) > > Tasks that contain other tasks (parralele/if/then/case/whatever). > > >Well, The requirement and how it is done are two separate things. I don't > >believe all tasks need to have access to their Task Object Model. > > agreed. > > >My idea > >of container tasks is that there will be a separate interface implemented > >by those tasks which will give the container the definitions of the > >contained tasks. > > precisely. > > >So, it may be required for some tasks, but not for all. > > So why did you -1 it? ;) In general, I don't think a task should know about its configuration info beyond its interface. It is up to the thing that contains the task to configure the task. I don't think the task itself should reach "outside" that boundary, if you know what I mean. Anyway, I don't think it is a requirement for all tasks to have this information. Whether container tasks need this information is not yet clear to me. Conor
