[linux-elitists] Of portable languages
Thu Sep 19 06:21:11 PDT 2002
On Tue, 2002-09-17 at 14:56, Justin F. Knotzke wrote:
> On Tue, Sep 17, 2002 at 11:42:11AM -0700, Nick Moffitt wrote:
> > begin Justin F. Knotzke quotation:
> > > The thing is, Freenet runs on more then just Unix.
> > So does Python.
> Which makes the original argument moot.
An interesting point that I've been meaning to discuss for a while.
When Java first came out, I went to one of Steel's talks and heard him
tout the benefits of Java's platform neutrality. At the time I thought,
"well, no it's strongly platform bound, but that platform is the JVM."
Still, it was an interesting tactic, and Java was not the first in the
game, but certainly the broadest experiment.
Looking back, I see a lot of problems with Java in the last several
years. Its not that Java cannot play ball with native conventions, but
that its adherents seem to be unwilling to do so. This poses a problem
because big and slow as many modern high-level toolkits may be (be they
for object sharing, GUI development, game design, etc), they are almost
always a better choice for performance than Java's native tools.
What's more, Java's desire to do things its own way leads to a certain
snobbery about local conventions. I see people write system tools in
Java and then wrap them in shell scripts in order to "integrate" into
the system. A Perl or Python programmer would almost never do this, as
they use the native conventions without a thought (which can make moving
a Perl program written for Windows to UNIX an interesting exercise in
culture-shock, but certainly not an insurmountable one).
So, in retrospect, I think Java was a wonderful experiment in platform
portability management, but ultimately it suffers from the cultural
under-development of its programmers, which it in turn fosters.
Languages like Python, Ruby, Perl, Scheme and other native-friendly
high-level languages seem to have the edge here.
Aaron Sherman <email@example.com>
More information about the linux-elitists