Path: utzoo!attcan!uunet!husc6!bloom-beacon!gatech!ncar!noao!ut-sally!nather
From: nather@ut-sally.UUCP (Ed Nather)
Newsgroups: comp.sys.ibm.pc
Subject: Re: GNU awk (gawk) on MS-DOS
Summary: MS-DOS gawk doesn't work
Keywords: awk gnu gawk
Message-ID: <11751@ut-sally.UUCP>
Date: 22 May 88 21:12:50 GMT
References: <1455@titan.camcon.uucp> <24837@bbn.COM>
Organization: U. Texas CS Dept., Austin, Texas
Lines: 43

In article <24837@bbn.COM>, kadler@bbn.com (Kyle Adler) writes:
> 
> I seem to have been luckier than some netters in that I've gotten the
> GAWK binary (as posted) to do some useful work.  The executable seems
> to work as advertised for string operations.  However, I've had no end
> of problems with numeric operations.  

So have I.  I got the book "The AWK Programming Language" by Aho et al.
and planned to find out what worked with Gawk and what didn't, and post
a summary to the net.  I didn't get very far: the first example in the
book, even when entered with double insted of single quotes, died a
horrid death with a "null pointer assignment" diagnostic.  It also
managed to clobber a TSR program in memory.

> For example, the simple script "{print i++}" results in seemingly
> random garbage, as though the machine is trying to print the ASCII
> value of the integer.  Forcing the variable into decimal format using
> "{printf("%3d\n", i++)}" seems to work for integers.  Unfortunately,
> even printf doesn't work properly for floating point results.  The
> script "{printf("%7.2f\n", NR/3)}" typically either prints zeroes or
> minus zeroes, or else hangs the machine.
> 

I found that *any* numeric value that I tried, using "print", would die,
although some of them would work using "printf."  I tried only integers.

> Has anyone else encountered these sorts of problems with GAWK under
> MS-DOS?  This program would be indispensable if only it worked.  I'm
> running it on an AT compatible.  By the way, I've gotten the version
> for UNIX to compile and run successfully.

Well, that's encouraging anyway, and suggests that perhaps the original
"port" to MS-DOS was either incomplete, or a partially-ported version was
inadvertantly posted instead of the working version.

I agree the program would be extremely valuable if it worked.  As is, it
just raises the frustration level to a very high value.

-- 
Ed Nather
Astronomy Dept, U of Texas @ Austin
{allegra,ihnp4}!{noao,ut-sally}!utastro!nather
nather@astro.AS.UTEXAS.EDU