“I have started a Users Group in the Harrisburg Area. I am having trouble locating users. The name of our Group is: Harrisburg Area Timex Sinclair Users Group. H.A.T.S. meets on the third Wednesday each month. I started the group with a few names I received from Russell Electronics. I originally visited the users group in State College, Pennsylvania, but that is too far to drive.”
Sincerely,
Harrisburg Area TS Users Group
c/o Dave Bennett
[address omitted]
Lemoyne, PA 17043
Editor: Hope that printing your address here will bring in some more members for your group. I might add that we will print the address of any group that sends in the information.
To the editor and the readers,
“I want to introduce myself to you. My name is Charles E. Goyette. I live in the province of Quebec (Canada) where French is the spoken language. That is the reason why my way to write is so strange; I am not used to English.
I liked Time Designs so much that I decided to give my programs to the readers. CAVERN, the program you have seen in the last edition (May/June ’86), was not, I admit, an excellent program. I had a few “bugs” and it was slow too. But in the month of May, I programmed the game SKI, which is listed elsewhere in this magazine. It is, I think, a pleasant game which has speed, and features good improvements compared to CAVERN. I hope you will enjoy it as much as I enjoyed programming it.
I must say on the other hand, that we have lost one brilliant programmer for the Timex computers…John Coffey [Editor’s note: he is programming for the Atari ST line of computers now]. I imagine that it took him over 300 hours of work to create DIAMOND MIKE II, a game that all of us should have bought…there is always someone “crying” because there is no software for the Timex computers…here is a guy who is just doing that: software! But John, I wish you good luck in all your enterprises. And to you the readers, it’s not too late to buy DIAMOND MIKE II. But don’t wait…”
Charles E. Goyette
Drummondville, Quebec
Canada
Editor: Charles, I really enjoyed playing SKI the other night–it’s addictive and really gave the flavor of down hill (Alpine) skiing (a favorite past-time during the winter months here in the northwest). One of our readers requested an explanation on how to program the little “spaceship” figure in the CAVERN listing (and probably the figures in SKI as well). Maybe you could enlighten us with a programming tip for the next issue?
I had a chance to speak briefly with John Coffey at the T/S Computerfest in Cincinnati. Although he praised the Atari ST’s, he mentioned how easy it was to get good results from his Timex/Sinclair. I have a “hunch” that we might be hearing from him again. Also note that Diamond Mike II is available from JRC Software, WMJ Data Systems, RMG Enterprises, and other dealers as well.
“I read with some interest, in the March/April ’86 issue of Time Designs, The problems James F. Brezina had with the fact that the function INT (SQ(675)) equaled 25. He felt there was something wrong with his 2068, and your reply stated that your 2068 did the same thing, and the “flaw” was in the ROM. In actuality there is no flaw. All computers use some sort of “successive approximation” to evaluate functions such as the square root. It carries the approximation to the point that successive answers are, to the accuracy of the computer, equal. If you will ask the 2068 to PRINT 26-SQ(676), you will find the answer is something like 7.45E-9. Since the computer is accurate to 8 figures, this small residual is treated as zero. That is, it is treated as zero by the computer but not by the INT function which truncates to the next lower whole number.
This is not a problem unique to the 2068. I have worked on very large computers, and had essentially the same problem. The lesson to be learned is that branch points should be tested before running a program. There are a number of ways to circumvent the problem. Mr. Brezina used one in changing the store number to 687. A second is to add .5 to the number for which the integer value is desired. This assures the number is rounded, and not truncated. A third method is to provide a range for the branch value. For example, if the branch test is for some function to equal 0, then the instruction should be for the absolute value of the function to be less than some small number such as .0001.
I hope this information will aid others who are writing programs to avoid a pitfall that took me over a week to discover the first time I ran across it. And reassure 2068 users that Mr. Brezina’s problem does not indicate a flaw in their machine.
Very truly yours,
Vance J. Carpenter
Fairport, NY
Editor: Perhaps “flaw” was a poor choice of words…thanks for setting the record straight, and sharing your tips on correcting 2068 mathematical routines–a subject that is briefly discussed in the users manual.
“As an additional bit of trivia on Richard Hurd‘s keyboard article, try a coating of clear fingernail polish to the keys. This will stop the wear-out of the letters. I have tried several brands and have found that Max Factor #13 Clear is the best. It leaves a clear transparent coating. I have two 2068’s…the first I wore the letters off the keys in nothing flat. The second one I coated, and it is still going strong after a years hard use. E. Arthur Brown has key legends with sticky backing for $1.50 a set. I got one and they seem to be real nice.”
P.S. Amen to Dunnington‘s “All Caps Please”, I am an old duffer with trifocals.
Dudly S. Rea
Priest River, Idaho
Editor: Mr. Rea also enclosed a circuit diagram for adding automatic two-key entry from a one-key closure, to answer a question posed in Richard Hurd’s external keyboard article (see the May/June issue). Ran out of space, but we’ll print it in the next issue.