From: Jim Jagielski <jim@jaguNET.com>
Message-Id: <199610021253.IAA22189@shado.jaguNET.com>
Subject: Re: Possible bugs in current version
To: new-httpd@hyperreal.com
Date: Wed, 2 Oct 1996 08:53:05 -0400 (EDT)
In-Reply-To: <199610021248.IAA18513@madhaus.utcs.utoronto.ca> from "Rasmus
 Lerdorf" at Oct 2, 96 08:48:22 am
Content-Type: text
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

Rasmus Lerdorf wrote:
> 
> >   *-linux*
> >   *-unixware*
> 
> I think to be safe we should set Linux to "no".  There are so many
> different distributions out there.  On my Linux box, "yes" works,
> but I am not very confident that this can be said for all Linux setups.
> 
> But Unixware!  A definite "no" there.  I had all sorts of problems with the
> regfree() call in the system regex library on a Unixware 2.0.3 box.  
> Another annoying thing is that in the Unixware /usr/include/regex.h file
> they define REG_STARTEND which to me would signify that the regex library
> supports the Start/End feature also supported by Henry Spencer's version.
> But nooo, that isn't the case at all.  They just define REG_STARTEND for the
> fun of it it seems.  
> 
> Anyway, I should be able to provide a more complete list of broken platforms
> soon.  My PHP/FI package went out with what was in effect a default setting
> of "yes" for all platforms for which configure found a regcomp() function.
> I am getting problem reports flowing in, and I will make a list of the
> systems on which I am able to verify that the problems are regex-based.
> 

A default of 'no' means NOT to use HS's supplied regex. I think you
want 'yes' for Linux and Unixware...

-- 
Jim Jagielski  << jim@jaguNET.com >>   |   "If we took the bones out
  **  jaguNET Access Services  **      |    then it wouldn't be crunchy"
      Email: info@jaguNET.com          |            Whizzo Crunchy Frog
++    http://www.jaguNET.com/         +++      Voice/Fax: 410-931-3157       ++