Some senseless babbling from Jonathan De Boyne Pollard to Lee Aroner
on 12-12-99  17:50 about Install...

 JdBP>> I think that Leonard is trying to make the point that using 
 JdBP>> FCBs for "fast deletions" is pointless in those 
 JdBP>> environments, because it *isn't* actually faster that doing 
 JdBP>> things the more up-to-date way using a simple 
 JdBP>> findfirst/findnext loop.
 
 LA> I *have* compared it Johnathan, and it's a whole big pile faster    
 LA> than running a loop, which, of course, is the backup routine in    
 LA> case the drive in question is HPFS (since warp intentionally    
 LA> traps if you try this on HPFS with a wildcard spec).

 JDBP> It's only faster on DOS, though.  It *isn't* faster on non-DOS
 JDBP> operating systems, such as OS/2 or Windows NT.  Try measuring it on
 JDBP> something other than DOS.

 JDBP> This was demonstrated a year or so ago in this very echo, with people
 JDBP> using the 4DOS DEL command with and without the /Q switch (which
 JDBP> selects between using FCBs and using MS-DOS version 2.0 style deletion)
 JDBP> and discovering that the two operate at exactly the same speed. 
 JDBP> Unfortunately, I don't remember who it was who sat down and tested it.

I can't say I was the only one, but I did post about running 4DOS's DEL
command with and without the /Q switch.

You were confused at the time about what it is that I was advocating,
because I did in fact report that 4DOS's deletes were substantially faster
with /Q than without.  I was not, of course, trying to say that FCB usage
was the reason, and in fact took some amount of time explaining that 4DOS's
/Q switch has nothing whatsoever to do with FCB's unless you're running
native DOS.

Under DOS, the /Q switch both suppresses output, and utilizes FCB's if the
deletion mask is suitable (not an extended 4DOS mask).  Under OS/2, the /Q
switch merely suppresses output.

When I first ran the test, it was on a 486DX2/66 machine, and the
difference in deletion speed was quite dramatic, between using /Q and not.

This was due entirely to the fact that screen I/O is expensive, and that
expense is readily apparent on a 486 system.

When I shortly thereafter upgraded to a Pentium/200, the difference became
quite negligible.

The discussion, incidentally, took place between two and three years ago.

I still remember writing about the brand new name of Chicago, complaining
that it wasn't even 1995 yet, in OS_DEBATE.

Mike Ruskai
thannymeister@yahoo.com


... Conscience gets a lot of credit that belongs to cold feet.

___ Blue Wave/QWK v2.20
--- Platinum Xpress/Win/Wildcat5! v3.0pr2
138/2
397/1