Wed, 03 Feb 1999 18:03:45 -0800
Billy Tanksley wrote:
> > y0,
> gR33t1ngZ! (Eww, I feel dirty now. :-)
wh!? !$ !7 n07 d3v!n3 70 $p33k 7#3 l!ng0 0F 7#3 h4x0r?
> > I had some ideas about the HLL... It should be a functional
> > programming
> > language or based on some mathematical scheme that incorperates forth
> > verbs and
> > a noun object extension.
> I love playing with language design.
:) go to my homepage and work on my lingo project!
> Second, IMO in the "testing" versus "formal proofs" battle, the
Not the point. Portability is. My theory is that if the program is described in
terms of mathematics it could be translated more effectively if not efficiently.
> > It shouldn't have any APIs in the language because that
> > would be too restricting and would make it obsolete too soon. Instead >>the program would have a set of bindings that would tell the system how >>to incorperate and present it.
> Talk some more. I almost understand that, and it sounds like it should be useful and achievable.
okay. though your following refferance is the same thing... The problem is how
to generate the "superclass". One possibility is to use natural language to make
an educated guess how to represent it... This is a totally diffrent type of
language problem. We are not trying to communicate a message bit a meaning and
pourpose... "this X is to save" where X can be any user action that ititiates
the save action... Not only user interfaces but other low level modules too!
Scenerio: Module foobar is installed on a system running my OS. The first thing
it will need to do is request from the parent module information about the
system. It will then recieve a list of apis available in the current evironment
and other sibling modules... Then the problem becomes one of binding/linking to
those apis and connecting to the sibling modules and establishing protocols with
them... CORBA uses "IDL" for this pourpose... Obviously this requires a great
deal of precision...
Think of all the kinds of storage media possible... (scanners, holo storage)
> Hmm. How does this relate to Geos' object model, where certain
> types of objects can be defined as having a 'variant' superclass, and the
> superclass for them is provided by the system at run time? (This allows,
> for example, TOTALLY different user interfaces, ranging from
> different-looking buttons to pen gestures in place of menus).
Probably the same idea. Just me reinventing the wheel again... :P
> > This would ideally run nearly optimally on all kinds of strange
> > and unusual hardware.
> How strange?
Quantum devices, FPGA. chaos processors, nanomechanical devices... u name it!!!
> > something of this nature:
> > bind ABC icon XYZ
> > bind DEF keystroke ALT-U
> > Icon XYZ 3dfractal NMO
> > bind TOR "howdy user!"
> Ah, I suspect this is what you meant by "binding script".
> Essentially what other systems might call a linker specifications file, huh?
guesso. Never forget I'm a moron who can't pass a college english course...
"We come in cubes!"
- The Borg