number recognition tools

[Andy Glew]

Urghh... 

I like showing people my source code, but I *hate* it when
they find embarassing errors (egoless programmer, I'm not).
Swallowing foot, let me acknowledge Joseph Pepin's bug report
(which he couldn't post) to the number recognition tools 
I posted recently.  I will posted a cleaner version soon.


Return-Path: <att!cbnewsl!joepepin at uxc.cso.uiuc.edu>
Date: Thu, 18 Jan 90 15:26:53 EST
From: jdp at tarpon.att.com (Joseph Pepin)
To: aglew at oberon.csg.uiuc.edu
Subject: Re: number recognition tools
Newsgroups: alt.sources
In-Reply-To: <AGLEW.90Jan15115309 at oberon.csg.uiuc.edu>
Organization: AT&T Bell Laboratories

Two serious bugs:

1) The shar file is corrupt: the END-OF-FILE after each man page
is preceeded by an "X", causing the next file not to be unshar'ed.
Maybe the original files were not terminated by a NL.

2) Executing "getnumber -100e-2" dumps core on a 3B2 and a 6386WGS
running SVR3.2.  The problem is you don't malloc enough space
for the right side of an infix expression:

Line 218 of number.c:

	right  = malloc ... InfixStrLen)));

should be:

	right = malloc ... InfixStrLen))+1);

I suspect that other systems treat malloc(0) as malloc(1).

Please fix these bugs and/or post this message: my newsserver
discourages postings.

--
Andy Glew, aglew at uiuc.edu