Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!seismo!rutgers!ames!ucbcad!ucbvax!FORSYTHE.STANFORD.EDU!JLORD%WATDCS.NETnorth From: JLORD%WATDCS.NETnorth@FORSYTHE.STANFORD.EDU Newsgroups: comp.sys.atari.st Subject: (none) Message-ID: <8704281930.AA25154@ucbvax.Berkeley.EDU> Date: Tue, 28-Apr-87 15:31:30 EDT Article-I.D.: ucbvax.8704281930.AA25154 Posted: Tue Apr 28 15:31:30 1987 Date-Received: Thu, 30-Apr-87 02:17:05 EDT Sender: daemon@ucbvax.BERKELEY.EDU Distribution: world Organization: The ARPA Internet Lines: 49 Received: by WATDCS (Mailer X1.23bUW) id 9745; Tue, 28 Apr 87 15:26:20 EDT Date: Tue, 28 Apr 87 15:24:53 EDT From: ORD JL (PHYSICS) To: INFO-ATARI16@SCORE.STANFORD.EDU Some user's comments on Fast (Cartridge) BASIC (by Computer Concepts in England) to supplement the review in May 87 Antic: 1. The review is misleading in one important respect. There IS a runtime package (in fact two of them) which enables programs to be run without the cartridge. I use the cartridge port as a 16-bit bidirectional interface for process control, and convert the operating program to a stand-alone .PRG file. 2. From the user's (as opposed to developer's) point of view Fast BASIC is a complete software system with an integrated assembler and documented keywords in place of (undocumented?) system calls. The assembler supports the full set of mnemonics (which is more than A Seka does), and the only feature I find annoying is the need for a BASIC loop to execute the 2 passes. There are a vast number of keywords, but they are all upper case, and you are warned that it is safer to use lower-case variables. 3. Fast BASIC is amazingly fast for an interpreter (and shows just how bad ST BASIC really is). I use the Infotek benchmark (created some time ago to test HP9830 calculators): FOR i=1 TO 100:k=SQR(INT(i+i~SIN(i))*LN(i)/ABS(i-COS(i))~SQR(i)):NEXT i (Only the LN is peculiar to Fast BASIC). The benchmark runs twice as fast in Fast BASIC as in ST BASIC, and I expected my optical analysis program to do the same, but it ran much faster, executing a 60-second ST BASIC calculation in only 13 seconds. The Infotek benchmark shows up one of ST BASIC's other problems - a progressive error in k beyond i=90: i k (ST BASIC) k (Fast BASIC) 90 1.34049E-08 1.340492E-8 95 7.60044E-09 5.512914E-9 100 7.06607E-09 2.240934E-9 4. Fast BASIC certainly has its share of problems. One unusual one I found is that it evaluates k=EXP(-x) as -1 when x=0, and as the review points out it does like to throw bombs up on the screen (making you glad that you have a bomb-proof ram disk). 5. Things to note when converting programs to Fast BASIC: (a) Fast BASIC tokenizes ASCII files on input, and if it runs into trouble it bombs - it is safest to insert \'s (for comment) into ST BASIC programs before loading in Fast BASIC and editing. (b) If you BLOAD and BSAVE floating-point arrays you will find that Fast BASIC and ST BASIC use different floating-point formats. 6. Programs in Fast BASIC can be written in PROCedure format with local variables and without the GOSUBs, line numbers etc. which so outrage computer scientists. They might even get to like it, but I suppose they will claim (correctly) that it is not really BASIC. ] Jack Ord, Physics Department University of Waterloo