Your Turn, Or I Get Letters

So I do get mail and I do appreciate hearing from you.

I received a couple of notes regarding language support for decimal datatypes. Here's one of my favorite letters, mainly because Jonathan calls me "Sir" and that is so infrequent around the office:

Dear Sir, The IBM 1620 computer had floating-point decimal in hardware.

http://www-03.ibm.com/ibm/history/exhibits/mainframe/mainframe_PP1620.html

I used to do linear programming on it in Tel Aviv in the 1960s.
Best regards,
Jonathan Rosenne

Thanks Jonathan.

I particularly appreciated Jim Keohane's letter, not because it had to do with COBOL, but because he called me "Mr. Erickson." That's almost as rare as being called "Sir".

Mr. Erickson, COBOL's only advantage is that it allows upper management to *THINK* they know how to program! Sprinkle some English-looking verbiage and use terms like sentence, paragraph then observe how truly frightening management can be. {smile} Cheers, - Jim p.s. The generally accepted COBOL programming productivity rate of 12-to-20 LOCs (lines of code) per day is what I term COBOL's ESCAPE VERBOSITY!

Thanks Jim, I'll be sure to point out to the boss the part about management.

On the processor-support side of things, William Boyle nicely filled in some gaps:

I believe that the 8087 chip family (including the Pentium and other derivatives that support the '87 instruction and register set) support BCD math in hardware. So, that being the case, the Power6 chip isn't the first to do so. Also, the '87 family will perform those computations in 80 bit extended precision. The problem is that there is little language support, as you have so rightfully pointed out. Some 3rd party BCD packages for C and C++ will support the native '87 code, but most just do it in software, which is not particularly quick. Unfortunately, I do not know off-hand for your readers which ones do support the 8087 BCD instructions. I considered doing this myself once-upon-a-time, but never got a round tuit. :-) Sincerely, William Boyle

While you and I don't know which onces support the 8087 BCD instructions, maybe someone else does. Please let us know. Also, you might look at the Dr. Dobb's interview with William Kahan who played a major role in the design of the 8087.

Now when was the last time you read anything about the Atari 800? I have to admit it has been a while, which makes me appreciate even more this letter from Paul Dubois:

I'm pretty sure the Atari 800 had decimal arithmetic in hardware. The reason I'm pretty sure is that it bit me. I was trying to learn to write a vertical-blank interrupt routine that recomputed the display and it seemed to work but every once in a while the screen would go nuts. I eventually tracked it down to having caught the interrupt some times with the FPU in decimal mode instead of binary, so all my calculations went nuts. If I saved the mode, put it into binary, and restored the mode before returning, all was well. I would be ashamed to admit how long it took me to discover this.

I had two disk drives, my own tape unit (!), and a modem that was probably no faster than smoke signals. I had cartidges with compilers for Basic and assembler. When the day came to throw it away, I had to ask my wife to do it; I just couldn't.
Paul Dubois
Contributing Editor
Computing in Science and Engineering

Paul went to provide a spec sheet on the from Oldcomputers.net that detailed the Atari 800.

Jonathan Wood's note deserves mention not for his first name, but because he rightly noticed my penchant for programming languages -- and he mentions once I hadn't run across:

I took a look at your site and noticed that you mention a lot about different types of programming languages. I would like to invite you to take a look at Aurora, which is a OOP style language. The Web URL is http://www.ionicwind.com. Paul, the creator of the language, has also released EBasic, which is a BASIC-style language. I was wondering if you would write up a review on either Aurora or EBasic or even both. Both development languages are alternatives to Microsoft's programming platform, and can run without any framework that Microsoft requires for there applications. Best regards, Jonathan Wood

Thanks Jonathan. I'm looking forward to playing around with both.

Then there was the note from Paul Clarke.

Hi Jon,
For all I know this could be the 6502nd comment you've received about your essay "Ada: Let's Hear for the Tried and Trusted," in which you report that the CPU used in a pacemaker is "the vertiable 6502." Before concluding that you had committed a simultaneous typo and malapropism I tried to learn whether "vertiable" is in fact a word. Every Google reference for "vertiable" that I examined looks like a typo for "veritable." I believe the intended word was "venerable," because lots of folks regard the 6502 as an elderly device worthy of love and respect.

Please forgive this trivial observation, but consider the possibility that for many people the 6502 is indeed venerable.
Best Regards,
Paul Clarke

Where's a good copy editor when I need one? But you're absolutely right Paul -- I meant "venerable" as in "the venerable editor goofs again." Consider me among the many people who find the 6502 venerable. (By the way, I did fix it.)

And as if that weren't enough, Jay Ballinger picked up on yet another of my goofs:

Hi Jon,
The title of your Pattern Language entry reads, "ISAS Revamps Web Site," but the association, which you reference correctly in your opening sentence, is titled, "IASA." Maybe we need a link to an international QA association? ;)
+jay

No Jay, what we need is a vacation. Thanks for pointing it out the error. It too has been fixed.

Thanks to you all, and keep those letters coming.

Posted by Jon Erickson at 03:51 PM  Permalink