At 05:49 PM 8/11/00 -0700, Madeline Schnapp wrote:
> >Perl is still struggling to find its ecological niche. You're not going to
> >write a device driver in it, or a windowing system, or an air traffic
> >control center. No-one would get in trouble for picking C++ for those, but
> >that's okay; Perl doesn't have to be all things to all people.
> >
> >At some stage, however, we oughtta acknowledge that a language with as many
> >features as Perl is too big to be considered only for programs of under 100
> >lines. Something with that much capability oughtta to be a candidate for
> >complex tasks. The prime niche here is enterprise-wide systems: product
> >management chains, document management systems, process control, CRM, ERP,
> >etc. These things don't have to bit-twiddle or require pedal-to-the-metal
> >optimization, they should be coded at the kind of '30,000 feet view' that
> >Perl gives you, and need to interface easily with HTTP, databases, XML, etc.
>
>There is a web-development framework written in Perl that has been
>developed by a private company (we are trying to convince them to release
>the product open source) that is over 100,000 lines of code and still
>growing. The program is feature rich (unfortunately not well documented at
>this point in time), as far as I can tell the fastest thing on the planet,
>and all around pretty amazing. Also, the Federal Reserve Board has some
>pretty hairy programs written in Perl. So, it can be used for large
>applications but is not the program of choice for large distributed
>programming environments for reasons that "Abigail" eloquently stated
>previously.
>
>A group in Agilent Technologies used Perl to develop their CADDATA Store
>web site written entirely in Perl and is pretty snappy (of course they have
>some big iron behind the site which helps of course).
>
>So Perl is used frequently in some very large applications but those apps
>are just not marketed very well. A killer web development framework just
>might be the "killer" app Perl needs.
Not that those aren't cool examples of large-scale uses of Perl that you
want to trumpet, but they are the exceptions that prove the rule. Two of
those three things are in-house products only useful to their creators; one
is an actual product. Compare that against the number of Java Beans and
other components flooding the market for doing major enterprise stuff, and
you see my point. If Perl has one product on its side then the ratio isn't
zero, but it's still pretty small. (Yes, I know there are others, e.g.
Lyris, but the count is still low.)
Neither should it be necessary for a product to be Open Source before it
can be marketed. Sure, that's my preference, but if a company wants to
make a product and release it some other way, Perl should be a choice for
them also. (But I concede that if people who can create that kind of
feature for Perl are ideologically opposed to it, we can hardly force them
into it.) Plenty of C programs are open source; the decision doesn't have
to be made by the language, people can make it on their own.
I'm not part of any group that stands to make money from selling binary
compiled Perl programs; I'm speaking purely from advocacy beliefs.
--
Peter Scott
Pacific Systems Design Technologies
