Re: gawk infinity issues

[Aharon Robbins]

> Date: Sat, 07 Jan 2006 11:47:18 -0500
> From: "Andrew J. Schorr" <address@hidden>
> Subject: Re: gawk infinity issues
>
> So it seems that every other modern version of awk treats those
> strings as numeric values.  Why should gawk be different?

You are making the same (easy to make) mistake as Paul made.  The "modern"
versions are doing nothing more than relying on the library atof/strtod
that understands the special strings.

Take those same awk versions and compile them on an older system and
they will behave differently.  Gawk does its best to behave the same on
all systems, without relying on the quirks / conformance level of the
underlying library routines.

It's all well and good to cite the current standards, but anyone who has
been in the GNU business for any length of time knows that the world is
a much less friendly place, and GNU software cannot rely upon standard
conforming systems to the last detail.  At least, I will not.  Nor,
for me, is gnulib a solution.

I am physically exhausted from a demanding day job and family life, and
do not feel like further language lawyering at the moment.  I am more open
to giving special treatment to nan and inf values than I am to hexadecimal
floating point, but that's all I care to say about it right now.

I will be unhappy if I am flooded with further emails on this issue.

I thank everyone for their input and for the proposed patches and will
give them all careful consideration when I can steal some time to give
gawk some attention.

Thanks,

Arnold