Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!usc!hacgate!ashtate!dbase!awd
From: awd@dbase.UUCP (Alastair Dallas)
Newsgroups: comp.databases
Subject: Re: dBASE IV again - a reply to a response
Message-ID: <166@dbase.UUCP>
Date: 18 Jul 89 16:17:36 GMT
References: <9467@xanth.cs.odu.edu> <154@dbase.UUCP> <9520@xanth.cs.odu.edu>
Organization: Ashton Tate Devlopment Center Glendale, Calif.
Lines: 51

In article <9520@xanth.cs.odu.edu>, aiko@cs.odu.edu (John K Hayes) writes:
> Can you tell me this - can you name ONE IMPROVEMENT IV made over III+????

In article <154@dbase.UUCP> awd@dbase.UUCP (Alastair Dallas) writes:
> A file that dBASE III PLUS takes 55 seconds to LIST takes only 22 seconds
> in dBASE IV.  That's ONE IMPROVEMENT.  I could go on, of course.

In article <9520@xanth.cs.odu.edu>, aiko@cs.odu.edu (John K Hayes) writes:
> BIG DEAL.  Who wants to list files to the screen so they fly past without
> being read?  If you look at useful processing tasks - such as simple BROWSE
> operations - IV takes about TWICE AS LONG as III+.

BROWSE on an indexed 25,000 record file takes less than a second to Page Down,
and approximately a second to Page Up (scroll backward).  dBASE IV is slightly
slower than III+ at this because IV's BROWSE is much more powerful and 
flexible.  IV's BROWSE has pull-down menus which support file navigation
without leaving BROWSE; you can BROWSE in one window while you do other
things in other windows, and so forth.  How important is 77/100th of a 
second versus 93/100ths in an operation that is user-bound anyway?

LIST is a more useful measure of a data engine's performance because it 
involves indexed record retrieval without much screen I/O overhead.  In
contrast, BROWSE has line-drawing overhead but also pre-fetches while 
waiting for keyboard input.  If you don't think that LIST is BIG DEAL,
how about SORT?  A 1,000-record file takes 5:77 seconds to SORT in III+,
but only 4:90 seconds in IV.

Ashton-Tate is very concerned about the performance of dBASE--we may be 
conceding "the fastest" to others for now, but we are committed to being
the fastest with the most-est.  That is, there may be other vendors offerring
a subset of dBASE IV that runs faster, but no one offers you a faster IV.
We have run hundreds of benchmarks on dBASE IV and when I tell you that
IV is considerably faster than III+, I know what I'm talking about.

By the way, these timings (and the previous numbers cited) were done by
me--ad hoc--on the Compaq 386 next to my VT100.  I'm not releasing
Ashton-Tate numbers, I'm performing research that John Hayes could do
almost as easily.  All I ask is that attacks on Ashton-Tate or dBASE
employ at least a modicum of scientific method.  If you say that FoxBASE+
sorts faster than dBASE, you won't get an argument from me.  If you
maintain that III+'s data engine is slower than IV's without providing
a shred of evidence, that's something different.

/alastair/


Disclaimer: Comments about what "Ashton-Tate" is concerned about and
committed to notwithstanding, these are my opinions only and should not
reflect upon my employer in any way.  When I say performance is important
to A-T, it is from the standpoint of a day-by-day observer of the company,
nothing more.