From: Charles Sandmann <sandmann AT clio DOT rice DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: profiling with DJGPP
Date: Sat, 12 Oct 2002 15:28:01 CDT
Organization: Rice University, Houston TX
Lines: 22
Message-ID: <3da885d1.sandmann@clio.rice.edu>
References: <2 DOT 7 DOT 9 DOT 191MP DOT H3U9EW AT pauzner DOT dnttm DOT ru> <2 DOT 7 DOT 9 DOT GKAP DOT H3VJYA AT pauzner DOT dnttm DOT ru>
NNTP-Posting-Host: clio.rice.edu
X-Trace: joe.rice.edu 1034455124 27279 128.42.105.3 (12 Oct 2002 20:38:44 GMT)
X-Complaints-To: abuse AT rice DOT edu
NNTP-Posting-Date: 12 Oct 2002 20:38:44 GMT
X-NewsEditor: ED-1.5.9
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

> >> 2) getenv heavily used by timer functions for timezone info
> >>     (TZ, TZDIR, etc. - seems things are stored in static
> >>      variables but probably not in all places?)

> > Only once: the time functions cache the values of environment
> > variables.  If your program changes the environment a lot (by calling
> > putenv or setenv, for example), then caching could be ineffective,
> > since each time the environment changes, time functions must see if
> > some of their variables changed as well, and take note.  This is so
> > that setting TZ to a different value is correctly handled.

> The profile shows - each tzset force calling tzsetwall
> (TZ variable is not set in my environment).

Have you tried setting TZ to see if the profile changes/improves?

A casual inspection of ctime.c (I'm not an expert) looks like to me that
at least the first comparison of lcl_is_set should be !=0 instead of >0 
to cache the case where TZ is not set.  You could try modifying ctime.c 
in such a manner to see if it improves performance (and still works).