Skip site navigation (1)Skip section navigation (2)
Date:      21 Nov 1996 14:10:08 +0000
From:      Paul Richards <p.richards@elsevier.co.uk>
To:        davidn@sdev.usn.blaze.net.au (David Nugent)
Cc:        msmith@atrad.adelaide.edu.au (Michael Smith), terry@lambert.org, roberto@keltia.freenix.fr, hackers@freebsd.org
Subject:   Re: Who needs Perl? We do!
Message-ID:  <57sp638rpb.fsf@tees.elsevier.co.uk>
In-Reply-To: davidn@sdev.usn.blaze.net.au's message of Thu, 21 Nov 1996 15:07:43 %2B1100
References:  <Mutt.19961120162842.davidn@sdev>	<199611210344.OAA10837@genesis.atrad.adelaide.edu.au>	<Mutt.19961121150743.davidn@sdev>

next in thread | previous in thread | raw e-mail | index | archive | help
davidn@sdev.usn.blaze.net.au (David Nugent) writes:

> Agreed. But there's also the distinction between "needed" and
> "desired". Certainly *I* regard Perl as an indispensible tool
> for what I do. I'm less certain that everyone else would regard
> it in the same light, which is why I still don't think it appropriate
> for a *base* distribution.

Hmm, following on from a previous post this is a kind of interesting
argument these days. I *never* use sed or awk because I've forgotten
how to do things in them quickly since I now always knock something
out quicker in perl. For anything more involved than a quick script
hack then I'd always choose perl anyway.

So, for me perl is needed and sed/awk (and even sh) could go
away. Perl does it all much better. (Implicitly invoke your own smiley
before getting wound up about this, though I am essentially serious, I
never use any of the above any more and it's been quite a few years
since I worked on a system that didn't have perl).

> The point is not really whether perl4 disappears or not (it *must* do
> so eventually - it is, as I said, old and unsupported) but whether
> perl5 is needed in the base distribution.

Well, I wasn't that keen on moving perl into the distribution at the
time. The few scripts we currently have could have been written in C
without much effort but we should either decide to rip perl out
completely or accept Perl5 as an integral part of the system and make
more use of it.

> > Other arguments that have been offered for the latter in previous
> > discussions; "Perl is too big" (size is relative, disk is cheap),
> > "Perl would be too hard to track" (contrib scheme should fix this).
> >
> > I'm still open to argument on this; I just haven't heard a counter
> > that holds up under scrutiny.
> 
> If all that was required for a proper perl5 distribution was the
> perl executable itself, I'd have no real argument. It is all of

The "uneeded" bloat is not required to run perl, it's just assumed to
be there by a lot of scripts. If our system scripts didn't need it we
wouldn't have to install it.

I've always felt that we should have a modular install mechanism like
Sun used to have in 4.1.3 (not sure about Solaris, never installed it
personally). Each module was self contained, there was a base system
and then you added things like the man module, which included all the
man binaries and the pages. We could have perl module, which included
perl and all the perl scripts. As long as the scripts are something
can be left out (which currently they could be) then that'd work. If
you could live without the adduser script etc then you wouldn't need
to install that module.

-- 
  Paul Richards. Originative Solutions Ltd.  (Netcraft Ltd. contractor)
  Elsevier Science TIS online journal project.
  Email: p.richards@elsevier.co.uk
  Phone: 0370 462071 (Mobile), +44 (0)1865 843155



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?57sp638rpb.fsf>