Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!seismo!rochester!ritcv!ccivax!rb
From: rb@ccivax.UUCP (rex ballard)
Newsgroups: net.lang.prolog
Subject: Re: Turbo Prolog
Message-ID: <504@ccivax.UUCP>
Date: Thu, 20-Mar-86 22:04:16 EST
Article-I.D.: ccivax.504
Posted: Thu Mar 20 22:04:16 1986
Date-Received: Sat, 22-Mar-86 22:25:04 EST
References: <1524@decwrl.DEC.COM>
Reply-To: rb@ccivax.UUCP (What's in a name ?)
Organization: CCI Telephony Systems Group,  Rochester NY
Lines: 29

In article <1524@decwrl.DEC.COM> vantreeck@logic.DEC writes:
>
>A recent article in EE Times covered Borland Intl.'s
>announcement of Turbo Pascal. An incremental compiler with
				  ^^^^^^^^^^^^^^^^^^^^
>a speed of 100KLIPS on an IBM PC and 300KLIPS on an IBM PC
>AT. Does anyone know what benchmark was used to obtain
>these numbers, e.g., naive_reverse? Or is this just some BS
>from Borland's marketing department? 
>
>-George

Incremental compilers cheat!:-)  They tokenize, resolve references,
and do about 80% of the actual work while you are typing in the program.
Really good ones can link a module change to existing modules.  There
are about 20 other good tricks that incremental compilers can use.
A more honest :-) quote might be bytes of Object created per second.

Of course, since there are only a few incremental compilers around,
any comparison is kind of silly.  The main advantage is the ability
to interactively compile/link/load/run your program almost as fast
as an interpreter can do it.  The main disadvantage is that to get
this blinding speed, you have to use the provided editor. Some of the
older incrementals (Basic-09 et. al.) had LINE editors (yeachh:-)).
If you used a screen editor and "Loaded" the source, the initial load
could take a while (still quite fast).  Basic is an easy language
for incremental compilation, just use calls instead of "tokens".
It will be interesting to see a good structured language in an incremental
form.