IMO, (memory leaks aside) the quality and availability of the widgets is
*the* most important factor to the success of DynAPI.  People want to make
cool sites quickly.  Programming with DynAPI is a lot better than rolling
your own portability library (or other libraries available), but it is still
hard and time consuming to get things working on all browsers.  An extensive
set of widgets would go a long way to improving development time and
quality.

That said, here are some thoughts regarding widgets:

1) I think it would be great if the DynAPI web site allowed user/developer
to add widgets to a central directory.  It would be awesome if the site
allowed users to upload, run and download submitted widgets.

2) There should be two tiers of widgets. Standard widgets that are part of
the official distribution and widgets submitted by anyone.  The standard set
of distributed widgets should be expanded beyond what is available now.

3) I find the FAQ's approach of throwing up its hands about widgets counter
productive.  I hope this is a short term time saver opposed to long term.
The FAQ also states that too many widgets would scare people.  I disagree.
I also think an official tier of distributed widgets and a directory of
other widgets would avoid any shock if and when there are 100's to choose
from.  Assuming that the widgets are well organized - the more the merrier.
(BTW, I've run into problems trying to make my widget run on a geo-cities
home page.  I think they are doing something wierd to prevent you from
deleting thier ads.)

4) The path (e.g. "dynapi.gui.button") should be sort of like a COM Prog ID.
I think Richard is more or less doing this by adding the developer name in
the path.  But the idea I'm getting at is that *any* widget should be
globally and uniquely identified by such a path and that once a widget is
"published" the path ID should be constant.  This way it will be easier to
share components and write widgets that depend on other widgets.  Right now,
it's hard to be sure what a widget really requires by looking at the code.
I don't think there's anything to "do" to achieve this except to explain it
in the docs and enforce (i.e. encourage people to use it that way) it when
it makes sense.

I'll end by saying I think the developers are doing some great work.  A
good, open cross browser API is desperately needed and you guys are getting
close.  Keep it up.

Ken


> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Pascal
> Sent: Friday, March 16, 2001 6:14 AM
> To: [EMAIL PROTECTED]
> Subject: RE: [Dynapi-Dev] Widget Development
>
>
> I think richard's site is great for that.. it already has a full
> collection
> of widgets available.. so if he's up for it :)
>
>
>
> Pascal Bestebroer ([EMAIL PROTECTED])
> Software ontwikkelaar
> Oberon Informatiesystemen b.v.
> http://www.oibv.com
>
> > -----Oorspronkelijk bericht-----
> > Van: [EMAIL PROTECTED]
> > [mailto:[EMAIL PROTECTED]]Namens Eytan
> > Heidingsfeld
> > Verzonden: vrijdag 16 maart 2001 11:13
> > Aan: Dynapi-Dev
> > Onderwerp: [Dynapi-Dev] Widget Development
> >
> >
> > I deliberately pick this list to send this msg. to I would
> > Like to recommend
> > that any widget code anyone has finish or half finished
> > should be uploaded
> > somewhere so everyone can look at them and fix/use them. Everyone here
> > starts developing things but then they do something else and
> > we have a lot
> > of half widgets (I myself am very guilty of that
> > windowmanager, borders and
> > more).
> > If we had some sort of page or site to submit too I'm sure if
> > someone would
> > need someone else's widget fix-up the code and resubmit it
> > there would be
> > many more widgets and easier for simple developers.
> >
> > 8an
>
>
> _______________________________________________
> Dynapi-Dev mailing list
> [EMAIL PROTECTED]
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev
>


_______________________________________________
Dynapi-Dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dynapi-dev

Reply via email to