On Wed, 2002-07-24 at 11:28, Patrick May wrote:
> What are the relative priorities of these requirements?

off hand i would say:
  1. multi-platform
  2. native integration (either direct or mimiced)
  3. easy to code
  4. speed and flexibiity

#3 of course will be taken care of by the GUtoIa API.
 
> I don't think it will ever be possible to 'put an end to the problem'.

watch me ;-)

> The problem is that interface code, whether it's html, gui, whatever,
> has it's own set of objects with their own hierarchies, etc.  Matching
> the interface relationships to your existing object relationships will
> always be a mess.

i believe my implementaion of using intermediary bindings will fix this
mess.

> This is where I like your x.widgitize() idea -- it's a great *pattern*
> for integrating the presentation code into your objects.  This pattern
> requires you to discard notions about the "separation of
> presentation", and instead follow rules such as "put functionality
> near the required data".  I'm just not sure if it makes a good
> *library*, b/c as I said before, x.widgetize() benefits from providing
> generalized application specific widgets.

i too find the x.widgetize() idea compelling, but also beleive you are
right. it does not ultimately make for a good way of dealing with the
problem. conversely, i think a seperation of concerns is far from a
notion to be discarded. in my work thus far on the project, i have found
proper use of SOC to be much more powerful then my original x.wigitize()
notions.

~transami