X
xyzzybill
Objective C was an attempt to bring Object Orientation to C. At the time if
it's incarnation, the prevailing thought was that OO meant not only objects,
but sending messages to them, probably because of the increasing popularity
of event-driven GUIs at the time. (Note that Smalltalk was another muddling
of the two orthogonal concepts).
(At least the above is how I think about the technology of that time).
Personally, I think C is a great place to implement OO "on top of " and
moreso for having the luxury of analyzing the weak aspects of the C++ object
model. I would opt for a departure "new" language though rather than being
100% backward compatible with C (one of C++'s Achille's heels).
Is OO in C's future? Who's for it? What are the minimal features to get
"adequate" OO capability in the language? (Feel free to add other relevant
thoughts also as this is just a brainstorming post).
Tony
I'll also point to DataDraw for OO in C (datadraw.sf.net). Code that
uses it runs faster than plain C, with productivity for many
applications that I feel exceed C++ and Java.
While most programmers will advise against a new language effort, I
say go for it. There's not much chance it will become popular.
Consider that there are hundreds, possibly thousands of programming
languages, and only a few that actually get much use. Still, there's
a chance, and you'll likely have fun and learn a lot along the way.
Besides, a true optimist will ignore all those odds, and only a true
optimists has any chance of actually improving the state of modern
programming for the rest of us.
I'm currently building what may be about my 10th effort at a new
computer language: 42, currently at sf.net/projects/graillang (then
unix name will change to l42). Like DataDraw backed C, it will be
faster than C, it will compile to hardware like behavioral System
Verilog, and will be user-extensible beyond any compiler I ever heard
of. Code reuse will also be extended into new territory, well beyond
class based inheritance. I also hope to enable software to run faster
on FPGA based reconfigurable computers than on Wintel machines, and
the vast majority of 42 will be written in 42 as libraries, even class
support.
Like I said, you have to be an optimist. In the case of 42, I think
it can actually happen. The goals sound like crazy ranting, but if
you want to talk on another thread (currently we talk on datadraw-
(e-mail address removed)), I can explain what's up in 42.