develooper Front page | perl.perl6.internals | Postings from July 2002

Re: cvs commit: parrot/languages/perl6 Makefile

Thread Previous | Thread Next
Nicholas Clark
July 20, 2002 13:52
Re: cvs commit: parrot/languages/perl6 Makefile
Message ID:
(this message is not meant to sound like a flame, although I fear it may come
across as one. sorry)

On Sat, Jul 20, 2002 at 10:16:03AM -0700, Brent Dax wrote:

> #    test:
> #   -	for t in t/compiler/*.t; do perl -I../../lib $$t; done
> #   +	for t in t/*/*.t; do perl -I../../lib $$t; done

> May I suggest that, at least in the long term, we avoid unportable shell
> scripts and such?  This sort of thing makes it quite difficult to port
> to other OSes (like Windows) and wins nothing over a standard test
> harness.

IIRC it even makes it hard to port between Unix variants - some makes
invoke $SHELL with flags to make it exit immediately if any command returns
a non-zero exit status, while others don't. The former makes it hard to
write meaningful shell pipelines.
(IIRC I've had to write make rules of the form sh -c 'what I really wanted'

Also, who says my $(PERL) is called perl? Or that it's the first perl in my
$PATH. At one point I had /usr/local/bin/perl as 5.00404, which is
intentionally going to trip up anything that doesn't remember that I invoked
it as /usr/bin/perl

Nicholas Clark
Even better than the real thing:

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About