Subject:
|
Re: Functional languages and/or RCX wrapper code
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 16 Nov 1999 00:12:52 GMT
|
Original-From:
|
Brian Connors <connorbd@yahoo.comSPAMLESS>
|
Viewed:
|
421 times
|
| |
|
|
> Someone (Vlad Dumitrescu(sp?), I believe) mentioned
> the possibility of writing a wrapper for the RCX
> serial protocol. Well, I've been taking a look at
> Kekoa's send.c, and as long as Kekoa's cool with it
> I'm trying to slice and dice it into an API called
> Seance (basically factoring out his comm functions
> from the rest of the program), this with an eye
> towards possibly reimplementing spirit.ocx as a
> platform-independent C library called Spectre. It's
> in
> the early stages, though, so don't expect usable any
> time soon. (However, if there's interest (==
> contributed routines), I'll be glad to give more
> details.)
Okay, here's the idea:
send.c is a pretty simple program in structure; there
are only about five functions in it, three of which
(rcx_send(), rcx_init(), and rcx_close() ) form the
logical core of the Seance API. The logical fourth
function (in theory, rcx_recieve() ) seems to be a
side effect of rcx_send(); given that the RCX isn't
likely to send anything interesting without having
been asked, I suspect this isn't the problem one might
think it is.
The Seance API, therefore, will be implementations of
the following functions, renamed: se_send(),
se_init(), se_close(). Spectre will be written over
these to keep source code oddities to a minimum. The
upshot of all this is that to implement a Seance
module for your platform, all you have to do is
rewrite the relevant functions in send.c. I've got
send.c partially factored already; I'll post a tarball
when I've got something that will build.
Now, as for Spectre...
-I've taken some looks through the spirit.ocx docs.
It's pretty big, actually; Dave Baum certainly didn't
write NQC overnight. It doesn't seem like it'd be too
difficult, though; the only concern is implementing
the "control structures".
-The impending release of the Scout SDK may sort of
change the problem domain; the issue here is whether
anyone cares that Spectre won't speak Scout.
-Finally, the big question: what in the world is
anyone actually going to do with Spectre when it's
completed?
Okay. First step, factor out send.c. Second step,
figure out what has to be done to make this work.
Any takers?
/Brian
=====
--
|
|
Message has 1 Reply:
 | | Re: Functional languages and/or RCX wrapper code
|
| (...) Hmm. No real idea, except say we can. Frankly, something that has been nagging at me lately is: what happens with all of this wealth of software that we've created when RCX 2.0 comes out? There is enough software out there to do really great (...) (25 years ago, 16-Nov-99, to lugnet.robotics)
|
2 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
