Offtopic: Re: Perl make depend made faster (fwd)

Michael Weller (eowmob@exp-math.uni-essen.de)
Thu, 19 Sep 1996 15:21:45 +0200 (MSZ)


Well, let me add on, that I'm also one of this people never ever using perl,
for these reasons:

On Thu, 19 Sep 1996, Herbert Wengatz wrote:

> Peter <peter@udgaard.isgtec.com> wrote:
> +> Herbert Wengatz wrote:

> +> installed perl and I don't expect to. Over anything I favor c-programs and
> +> I never make anything else myself (including scripts).
> That's up to you.

I won't go that far, that I never use scripts. However, as a researcher, I'm
often travelling through the world using computers at zillions of different
universities. This teaches you to use basic tools. In addition, I
occasionally administer unix nets. You also only have basic tools in many
admin situations. Thus:

> But try to port your c-program to the following machines and OS's:
>
> DEC-Ultrix,
> Sun-Solaris,
> Sun-SunOS,
> PC-SCO-Unix - and worst of it all:
> PC-SCO-Xenix

Well, you may not need to port a perl script to any of these machines...
However, you have to port perl!!! Is it actually possible to port perl5 to
Xenix? With all features of it usable? I can't really believe that. (But
might be wrong). Apart from that, when these are not machines directly
administered by you: will there be a perl installed? Of the right version?

Probably they don't have it, they use a different (incompatible, but also
called /usr/bin/perl) version for some old scripts they use regularly,
you cannot get a perl source from somewhere that compiles out of the box.

Thus: You'll endup porting perl5 for the ten-th time for some strange machine
and install it on that account eating much memory from your low quota
guest account.

For a C-program -- if you write it plain K & R or very basic plain ANSI and
portable -- it will almost all times work out of the box, maybe with a few
pointless (!) warnings about signed and unsigned char, or void * for some
libc calls..

I myself would actually prefer awk here. Not because it is good, but
because it is a standard, easy to use and available on ANY unix machine in
the world, right out of the box. (again, I dunno about Xenix).

BTW, the same applies to vi<->emacs, sh<->csh (or bash or tcsh).

>From my own experience, it confuse me more to use two editors at the same
time (I'll always enter key sequences from one editor into the other; I was
very happy when linux appeared and I could also use vi (and unix) at home))
than I benefit from the powers of the mightier one where it is available;
same for perl/awk. Your mileage may (and obviously does) vary.

But again, all this does NOT apply here. The linux kernel make will never
be called on Xenix, so what? When the cc of Xenix can't compile Linus' tiny
elementar C proggy it will certainly not compile the kernel.

So, please stop this pointless, religous discussion on this list.

I must admit though, that I'd also not like to be forced to loose a few megs
for perl, tcl/tk, or emacs just to be able to compile the kernel. I certainly
have enough disks on my home machine for that, but still.. just install
perl for 'make depend' is IMHO indeed just not worth it.

> My motto is "use the right tool for your task!".
> You won't try programming a video-game in Cobol and a heterogenous distributed
> database in Assembler, don't you ?

Now, this is a very good motto! And I'm a friend of the same philosophy
too. Still, if I had a tiny proggy to code and knew it would be perfect
for lisp or oberon or whatever, I still won't go install and learn lisp
just for this single program and then never again use it.

Just my 2 cents, no pun intended,

Michael.

(eowmob@exp-math.uni-essen.de or eowmob@pollux.exp-math.uni-essen.de
Please do not use my vm or de0hrz1a accounts anymore. In case of real
problems reaching me try mat42b@spi.power.uni-essen.de instead.)