ac> Thanks for the FREQLOG.C source.

No worries.  It's often easier to write these things yourself
than hunt down a utility to do it (although you might want to
keep an eye out for "Binkley Log Analyser", which comes with
C source.  However, freqlog still took about 1-2 hours to get
working properly.  Mainly because I originally didn't have 
these lines...

>         if (*buf == '>') continue;
>         pbuf = buf + 21;

and was not masking out the debug info.  I didn't think of that
at the time, but had noticed that all my copious debug info 
seemed to start with lower-case, so I was using that.  Turned out
there was some rare debug info that didn't have lower case, so I
put in an exception for that, and it only got worse.  I then saw
the forest through the trees, and it became a lot nicer.

ac> I had to change this to

ac>     pbuf = pbuf + 23;

ac> though, for it to work with the logfiles I use here.  I guess your BTPE 
ac> uses a different log format than BTJD.

+ 30 Aug 00:00:13 BINK begin, BinkleyTerm-OS/2 Ver. 2.59 -IBMC/2

012345678901234567890123
0         1         2  2

+ 02 Aug 22:30:21.41 begin, BinkleyTerm PE OS/2 3.04 alpha -watcom

Interesting.  The one change I made to Binkley's log was to record
down to the 1/100 second level so that I could see what was 
responsible for the various delays.  I thought I had managed to
do an exact replacement in size, but obviously not.  I will
correct this if I put out another version of BTPE.  BFN.  Paul.
@EOT:

---