🚀 go-pugleaf

Have to add a few more "time man csh > /dev/null" timings and point out that
this is a rather invalid way of getting times to do something because these
machines use VERY different ways of getting the man pages (especially
the IBM RS/6000).

on:

IBM RS/6000 Model 220	40.44 seconds (first time)
IBM RS/6000 Model 220	9.2 seconds (second time)

ESIX SVr4, 386/40	2.96 seconds (first time)
Esix SVr4, 386/40	1.25 seconds (second time)

Note that on the IBM, it searches it's INFO Explorer stuff FIRST and then
resorts to /usr/share/man.  I agree that people should be testing the nroff
on the man page and not the man command.  I hear MASSIVE disk activity
when I do the man on the RS/6000.  But when I directly use info to look up
csh, it takes no time at all (I'm NOT using the CD-ROM info).  Of course,
on the 386, the man pages are already catman'd, so its an invalid comparison.
(The RS/6000 is roughly three times as fast as the 386, and six times as fast
on disk I/O).

-Nathan Lane
Trico Title Company, Santa Barbara, CA
nlane@well.sf.ca.us