Re: [Autogen-users] Use of ctype(3) functions

[Harlan S. <st...@nt...>]

I had similar issues with ctype(3) back in the 90s with NTP.

I very carefully made sure to pre-check all data sent to ctype(3) calls
to make sure they were in the proper "domain".  This was, of course,
back in the K&R days.

As time passed, and as POSIX and related standards came in to play, the
rules seemed to have tightened up, and various folks have "cleaned" the
ctype(3) calls throughout the codebase.  This concerns me, but we
haven't had any bug reports yet.

I'm also wondering what the circumstances might be that would trigger
any potential problems here.

Our QA team will be writing test cases for ctype(3), to make sure it
handles these things properly.

Oh, many years' ago I filed a bug report with gcc to get them to emit a
warning when somebody was using a signed char as an array offset.  It
took a little back-and-forth to make it happen, but the warning was
added.
-- 
Harlan Stenn <st...@nt...>
http://networktimefoundation.org - be a member!