Interface(s), etc. (was Re: release

Francois-Rene Rideau
Tue, 6 Jun 95 17:33:53 MET DST

> First off, I'd be dying to know if this here interface(s) project is 
> supposed to have a trailing s.
> It's more all-encompasing that way, but it makes it a bit akward to say.
Well, *you*'re the project leader, so it's up to you to decide. Currently
the directory in the WWW pages is named "Interfaces", but you can change it if
you want...

> One question I've been wanting to sort of throw out here is how it's 
> possible to develop the high-level items such as HLL, Interfaces, etc. 
> when there is no low-level stuff to work with.
   Well, we sort of have to: the LLL is meant to allow the most efficient
possible portable implementation of the HLL; only by refining the needs
of the HLL can we successfully write the LLL.
   As for existing LLL code, we could have some running platform (able of
computations, basic I/O, and trivial forth-like parsing et al) in less than
one man-week worth of work. Once I finish my exams, I'll focus on that.

> How can we design the HLL 
> syntax or some cool GUI without a foundation to test/run it on, without 
> test applications/objects being built alongside it?
   We must begin design without testing, and then build our own tools to test
as we continue designing. That's the hard way, the way frontiersmen must do.
   Unless we found existing tools that satisfy us, and thus realize that we're
not on the frontier anymore.
   Has anyone had a look at Oberon, SELF, or Dylan and could produce a
critique, and/or point to a way of making one's opinion (papers, ftp sites,
etc) ?

> As a high-level 
> person here, I'm feeling there are a lot of cool ideas to play with, but 
> I'm not certain where I'm able to start without having to write lots of 
> crappy, temporary low-level items that will just get thrown away in the 
> long run.  Any ideas?
   I guess we'll have to play with such crappy things to have things going.
If you prefer to wait for other subprojects to bring you tools, perhaps you
could meanwhile have a look at all the existing interface software in the
world (for instance, the Tk toolkit for X seems popular, and many languages
like Python, Perl, Icon(?), seem to use it; perhaps it's worth a look ?).
   The OTOP project strives to implement TUNES over POSIX; such knowledge of
X toolkits could be of direct utility in OTOP...