[bitc-dev] White space
wren ng thornton
wren at freegeek.org
Mon Aug 9 01:45:30 PDT 2010
Ben Kloosterman wrote:
> >Yes, BitC should try to go gently on its intended audience. I could be
> >wrong, but I don't get the impression that audience is programmers
> >who've never moved beyond C++.
> The vast majority of developers I have met are experienced in only a single
> language ( though they may know a few from Unie ) and unwilling to change.
> I suppose the real Q is whether BitC is focusing on challenging Erlang and
> other functional languages in the system/embedded space or is it looking to
> convert C/C++ devs ? C , C# , Java and C++ derivatives are probably 50-60%
> of programming and most of the rest is scripting PHP , Python etc. C and
> C++ are in need of replacement, it seems to leave a huge audience if you
> just target functional System programmers.
> A language making 1 % is no big deal as they say in that thread Cobol still
> has more ,and if you rise up to 1% and then start falling I don't think the
> language can be perceived as successful.
The reason there are so many Cobol jobs is because of legacy codebases
that need maintaining. I'm not aware of any companies doing active
development in Cobol. There's nothing special about that position; there
are very few companies doing active development in PHP anymore, but
there are still a lot of PHP jobs because companies need to maintain the
software that was developed during the big PHP craze.
The degree to which programmers try new languages depends predominantly
on where they work in the field. In the web development sector it's
commonplace to completely change languages multiple times in one's
career. Perl, Python, PHP, Ruby,... That's not because web devs are any
smarter than other 9-to-5ers. It's because that's the accepted culture
of web development. Any individual project is likely to be written in at
least three languages, not including HTML and CSS in all their versions
The multilingualism is the same in system administration, because of the
countless scripting languages needed to wire things together, configure
servers, and keep the network running. It's the same in natural language
processing, which is why I helped start the Joshua project, to try to
fix that. Joshua is all Java, but what it's replacing is in Bash, Perl,
Python, and C++, so all of the developers on the team must know at least
three of those five well enough to understand the performance hacks and
defensive coding people've come up with over the years.
The only developers I've met who were only experienced in one language
are now retired. I'm not saying they aren't out there, but they're
certainly not in any of the fields I've worked in. Which also includes
WSNs, robotics, and compilers--- so it's not anything special about
higher-level programming. In any case, I think this thread has ceased to
be productive. We've no need to rehash the language wars of the 1990s.
We all agree that C needs replacing, the only question is what the
people currently using C are like. To answer that would require polling
organizations currently using C, not debating on an esoteric language's
mailing list. While we're at it we should finally replace Fortran too.
More information about the bitc-dev