[perl6-language trimmed again]
At 05:08 PM 8/4/00 -0400, Chaim Frenkel wrote:
>You would need two (or more) perl executables.
Well, I wouldn't, but you're talking about the places that chose to make
their perl have strict by default? Let's try make up a hypothetical case.
Let's say XYZBank Corp. is a major house with a generous attitude towards
the use of Perl in financial applications, nay, even mission-critical
ones. Constantly hiring new programmers, quality control is a big issue as
this business expands. They want to remove as many barriers as possible to
bug squashing. Educating their programmer population is one approach, but
in really big institutions, education has its limits; corporate-wide
communication is a huge problem.
The key to understanding XYZBank Corp. is to realize that they don't have
one programming department, one QA department, one operations department;
they have dozens, all independent of each other. But nevertheless sharing
code with each other.
A manager (horribly overworked) of one of those operations departments
(horribly understaffed) wants to do everything possible to ensure that the
code he/she puts into operations is safe. Yes, they could check that the
code starts with -w and use strict; but if their perl has those by default
this person will feel even safer. Maybe more than justified. But it all
counts.
>The downloaded CPAN (or other source) scripts/modules that don't follow
>the conventions would either have to be ported to the local environment.
>Or they would need their own executable that doesn't follow the standard.
>
>One or more groups will suffer.
One can also think of it as encouraging good code. It's been a p5p tenet
that the core should be -w clean; if it's good for the core, why not
CPAN? CPAN quality control has been a sore point for a while. The places
that have their strict turned on by default will likely be of a mindset
that they would not want to use code that didn't work. When I compile
freeware, I get nervous if gcc spits out any warnings, unless the README
tells me they're okay. I figure that if the programmer didn't fix those,
what else didn't they fix?
On a completely different tack, think of it as saving typing. Look at
Randal's columns; every listing starts with -w and strict. Many people
express the view that they're essential for anything more than 20
lines. Why make all these people type that in every program if you can
make it the default? Why send, however unwittingly, the message that Perl
developers consider programs under 20 lines more important than the larger
ones?
I think I've said all I have to say on this religious issue... no need to
repeat myself after this. I'd like to see the -z idea mentioned in the
final RFC.
--
Peter Scott
Pacific Systems Design Technologies
