On Wed, 5 May 2004, Michiel Meeuwissen wrote:

> Johannes Verelst <[EMAIL PROTECTED]> wrote:
> > Any comments?
>
> It's probably a good idea. The only remark that I would like to make is that
> the possible values of a certain field, are actually not related to
> editwizards. The field-types project was a.o. started to be able to define such a
> thing more generally. If e.g. the possible values of one field follow from
> the result of some query, I'd say that you should define that query inside
> the definition of the field itself, somehow. That would of course also imply
> changes to dove and editwizard, because the concept of 'possible values'
> must be transferable by the dove-protocol.

While I agree that this would be the most elegant solution, I don't think
it's workable. Sometimes the possible values for a field depend on a lot
of outside variables.

For example; say you want to define newstypes within a site: for the field
value 'type' of a the 'news' builder you will need to know the site this
newsitem is related to. When asking a 'Field' for possible values, this
class needs to know about the context it is called from.

> When this is finished, all optionslist can be removed from editwizards,
> because everything is automaticly determined by the field's specialization
> definition. It will then of course also be available to other editors, like
> taglib-editors or so.

I disagree, since it might be very well possible to have a specific
editwizard where a subset-optionlist is shown.

> So, my hope is that every change on this field might soon become
> deprecated... We do need a good inventory of what kind of lists of possible
> values there are, becasue those need be 'definable', preferabbly without to
> much of java-coding. This will also imply other constraints like minvalue
> and maxvalue for floats and so on (The possibilities are endless so the
> trick will be to find something reasonably simpl, yet feasible enough to
> cover most things one might want).

I think it's a very good idea to put more logic into fields, but I don't
think that everything can be done there. When for instance possible values
depend on a context, this creates a lot of overhead in the Field
implementation. Putting it inside an optionlist-query is not such a bad
solution in my opinion.

Johannes
-- 
[EMAIL PROTECTED]   | It is always possible to agglutinate multiple
[EMAIL PROTECTED] | seperate problems into a single complex inter-
PGP ID: 0xFED127BD     | dependent solution. In most cases this is a
Tel: 035-6474202       | bad idea. (RFC 1925, Truth 5)

Reply via email to