Tuomo Valkonen wrote:
On 2006-05-15 19:36 -0400, Geoffrey Alan Washburn wrote:
Sascha Silbe wrote:
Infact, if I wrote this Vis paper now, I'd probably try do the example
in Haskell with static typing :).
Actually, I'm doing the prototype in Python.
Learn Haskell; it will do you a world of good. If you really want
to use a dynamically typed language, use Scheme, it can do everything
Python can, better, has many high performance implementations, and was
designed by people who actually knew what they were doing.
Well, I don't think the core of Vis should be written in Haskell, but
the examples in an article just as well might.
Looking at your Vis paper again, I guess it really depends upon what
you mean by "framework". I can imagine that there could really be
several different ways of realizing Vis:
(1) Vis as a "display" daemon.
(2) Vis as a library API (many different implementations, dynamically
linkable)
(3) Vis as an interface compiler (stylesheets -> linkable code)
I think as far as I can tell from the document you seemed to be
intending (2). In which, case I expect that there will be many
different implementations, and it wouldn't really matter what language
it was written in as long as it follows the system dynamic linking
conventions.
For things like rendering, etc. I agree the best
performance/portability would be obtained from C, but depending upon how
rich the stylesheet language becomes (and I can imagine that a
declarative constraint language would make a lot of sense here) it seems
like there would be a significant benefit to implementing that kind of
code in a statically typed functional language. Which I guess raises
the question of whether it makes sense to define the stylesheet language
alongside the Vis API, or whether to just have per UI stylesheet
languages (ui-ion's stylesheet language would be Lua, and ui-html's
stylesheet language could be CSS+JavaScript).
