Re: Strings and reading from a file

[Hyman Rosen <hymie@prolifics.com>]

Marin David Condic <condicma@bogon.pwfl.com> writes:
> However - enamoured as I am with the "perfect" solution that works for
> any case that may arise - I'd wonder if in practice it isn't a bit of
> overkill.

No, it's not. One of the most annoying things about the old, pre-GNU
UNIX utilities were the plethora of built-in line limitations they
had. It is in the exceptional situations where you most need your
tools working properly. If the tools silently discard lines which
are too long, or divide them into several lines, or crash, or die
with an exception, that's a very bad thing.

For example, my .newsrc file, which is a plain text file, has some
very long lines in it. I have a little awk script which finds the
widest line in a file. On Solaris, nawk aborts with

	nawk: input record `comp.lang.c++! 63853...' too long

On Linux, everything runs fine, and I discover that the longest line,
for comp.lang.c++, is 27530 characters long!

If you find yourself fighting the language in order to handle input of
arbitrary size, then it's a problem of the language. I know that it's
not much of an issue in C++, and I would guess it's not much of an
issue in Ada either. Limiting input to the size of a fixed buffer is
just laziness or incompetence.