New firmware for the Swissmicros DM15L

[Helix]

(Yesterday 07:47 AM)jebedeo Wrote:  As you might notice the firmware is lighter than before, there are some extra 4kB available now!

As you kindly asked us to provide some ideas for filling the free space ( ), here are mine:

1. Numbers with positive exponent are somewhat difficult to read, because the E forms a continuity between the mantissa and the exponent. There isn’t such a problem with negative exponents, because the minus sign breaks this continuity, and the presence of an exponent is obvious.

I’ve checked many vintage calculators and pocket computers from Sharp, TI and Casio, and all of them adopt one or several of the following methods:
- displaying a space between the mantissa and the exponent (in place of the minus sign for negative exponents)
- displaying a + sign in front of the exponent
- displaying a smaller E than the regular one (an easy alternative for your firmware could be the use of the lowercase e)

I would appreciate if you could implement one of these principles to emphasize the presence of an exponent.


2. Contrary to the HP 15C, there is no digit separator in the mantissa. This can be problematic with big numbers. The solution is to use the SCI or the ENG notation, but this means addressing the point 1. In this case, I don’t think that digits separators would be necessary.


3. What about programs displaying text messages? Of course, there is no alphabetic keyboard, so I imagine a special instruction followed by a row of ASCII codes, that would display the message in the upper line when the program stops or pauses.
That would be fantastic!


4. I will never complain if you improve the integrate algorithm
But this my last point on purpose, because I consider previous points more important.

[jebedeo]

Here's another beta with:

1. sin(180) is fixed properly it only took implementing sinpi() from scratch lol
2. With the new sinpi() available, I was able to fix the behavior of the complex power function cpow(), the one behind Y^X. It now handles cubes of negatives correctly and it fixes the bug that John found as well.
3. There is now a space between mantissa and the exponent, improving readability. Thanks for pointing that out Jean-Charles!

Quote:1E-323 ENTER 2 / gives 0E-315

This one has to do with subnormal numbers and is not fixed. My own naive implementation of printf is getting too complex to debug and I will re-implement it by copying what people did in the actual printf source code, so this will be fixed then.

Quote:What about programs displaying text messages?

That sounds interesting, I added it to the list and once everything else is done I might give it a go if there is room left

Quote:Contrary to the HP 15C, there is no digit separator in the mantissa.

Added to the list as well, thanks!

[Johnp_g]

Nice job!

It's a little faster too.

8-Queens is now complete in 1.222 seconds
Savage Benchmark is now complete in 4.22 seconds

I've tried the new Swiss Micros V33 firmware, but keep coming back to this one, as I find the permanent complex functionality and display of real & imaginary parts too good to lose.

And the speed is pretty amazing too (apart from the few odd integrate cases that are very slow compared to the DM15L stock firmware).

Cheers,

John