Xref: utzoo comp.sources.bugs:814 comp.lang.c:7621
Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!decwrl!decvax!ucbvax!pasteur!ames!hao!gatech!udel!rochester!ritcv!cci632!ccicpg!nick
From: nick@ccicpg.UUCP (Nick Crossley)
Newsgroups: comp.sources.bugs,comp.lang.c
Subject: Bugs in 'check' program
Message-ID: <12799@ccicpg.UUCP>
Date: 26 Feb 88 20:19:20 GMT
Reply-To: nick@ccicpg.UUCP (Nick Crossley)
Followup-To: comp.sources.bugs
Organization: CCI CPG, Irvine CA
Lines: 21
Keywords: check,lint
Summary: 2 minor problems and one silly bug

A few errors have been pointed out in my recently-posted 'check' program.
First, the shar I used was not the most portable, and you might have to edit
the tests on the line counts after each extract to use a numeric comparison
(-ne) instead of a string comparison (!=).  Thanks for noticing this to
ccicpg!uunet!rlgvax!dennis (Dennis.Bednar)

Second, the lexical analyser does not allow form-feeds in the source.  Change
line 40 of lex.l to add \f (and possibly \r) to the set of 'white-space'
characters.  Thanks to uunet!amdahl!nsc!taux01!taux02.taux01.UUCP!yuval
(Gideon Yuval) for this one.

Finally, my entry for 'silliest bug of the year' competition.  In the routine
'cwarn' in lex.l, I declare on on-stack (auto) char[120], and then about three
lines later, I call free on it!  This does not make the next malloc very happy.
Blame late-night last minute changes before posting!  This was noticed by
uunet!ubvax!pacbell!ptsfa!fdp (Frank Pedroni)

Nick Crossley
CCI, 9801 Muirlands, Irvine, CA 92718-2521, USA
Tel. (714) 458-7282
uucp: ...!uunet!ccicpg!nick