Wizards and W3 integration (was: Re: A road map for Oort Gnus)

[Per Abrahamsen]

Lars Magne Ingebrigtsen <address@hidden> writes:

> Per Abrahamsen <address@hidden> writes:
> 
> > It would need some more work to be really useful, but the real problem
> > is that w3 is not integrated in Emacs.  WM Perry hasn't time to do
> > more than minimal maintainence, and nobody else have volunteered to
> > take over w3 or w3 integration.
> 
> I think that w3 is an essential Emacs component. 

That was my standpoint 7 years ago, and still is.

> These days, most information is out there on the Web, and to access
> that, we need the URL library and the HTML parsing functions.  In
> short, w3.  It's going to become even more vital in the future, what
> with all these XML-RPC/SOAP thingies that are rearing their heads.
> 
> Are anybody at the FSF dealing with these issues?

If "wishing someone would take up the project" counts as dealing, then
yes.

> It's not impossible that funding could be found for such a project,
> if that's the issue.  (I suspect that it's more a question of time.)

I think so too.  You need someone who is smart enough to understand
both Emacs and W3, yet stupid enough to volunteer.  

Ops, did I say stupid?  I meant, eh, "have time enough to spare".

> > I think using w3 would be the ultimate solution, but we could invent
> > an ad-hoc markup language for the purpose.
> 
> I think that would be counter-productive.  w3 is the way.  

I agree.  However, that notion have delayed wizards for 7 years now.

> Who do we bug to get things set in motion?  (I've Cc'd this to
> Bill.)

I have CC'ed it to the Emacs developers list, both Setup Wizards and
W3 integration are big questions that affect all of Emacs, not just
Gnus.

> > Depending on how much control is needed, it could be build upon either
> > custom or the widget library.  I don't think the programming part will
> > be that hard, the real challenge would be designing the wizards
> > themselves. 
> 
> If we had a proper framework for building wizards, I think writing the
> wizards themselves would be quite a lot of typing; yes, but not very
> complicated typing.

It is not the typing that is hard, but the design.  What are the
really important questions for this application?  Which things can be
safely assume?  How can we prensent the concept in a way the average
user can understand?

> I envision a kind of rule-based framework.  "The user has input this
> and this information, and this and this exists on the system, so we
> present the user with this information and these choices."
> 
> I don't think writing that kind of framework would be trivial...

I envision a markup language with embedded widgets and embedded Lisp.
Trivial stuff to implement, in fact, Bill Perry have already
implemented it (not that W3 is trivial, but this part is).  It just
needs to be integrated, and an example wizard.  As we get practical
experience from writing wizards, we would write more widgets and more
Lisp functions, making the progress evolutionary.