New HP-41 Module released: The Total-Rekall

[Ángel Martin]

The definitive "Total_Rekall" now available.

This one closes the circle somehow on all previous RCL math and stack exchange implementation.
Now with indirect and stack arguments supported - the full 9 yards.

It also adds the FIX_ALL functionality included in the SandMath - some users wanted it as a separate version. As added bonus I included the full UMS and constant library - perfect complement to fill up the available space not taken by the RCL/STK functions.

Soon at a TOS near you...

[Sylvain Cote]

Hello Angel,
Great! It is now available for download.
QRG: there are register name errors in the description column of line 14 to 19.
Thank you!
Best regards,
Sylvain

[Ángel Martin]

(04-29-2015 04:24 PM)Sylvain Cote Wrote:  QRG: there are register name errors in the description column of line 14 to 19.

thanks for the feedback - yes, the descriptions are obviously incorrect, to much copy/paste. I'm preparing a writeup with additional details, will have this corrected too.

Cheers,
ÁM

[Garth Wilson]

Wow, another home run!! It looks like it will be a great benefit in making programs both faster and shorter.

[Ángel Martin]

Here's the new QRG adding some details to the function table.

PS. The Floating FIX (aka FIX ALL) is also included on the function set - for those still not using the SandMath on a daily basis ;-)

[Dieter]

(05-01-2015 02:11 PM)Ángel Martin Wrote:  Here's the new QRG adding some details to the function table.

Ah, great – so now there's RCL arithmetics for the 41, at last. As well as a number of new tests. Here the QRG says that ?X> is a "greater than X" test, or ?0< is a "less than zero" test. However, the command name says exactly the opposite: if 0<X then X is greater than zero, not less. This would mean that for instance ?X< 01 actually tests if X > R01. ?!?

(05-01-2015 02:11 PM)Ángel Martin Wrote:  PS. The Floating FIX (aka FIX ALL) is also included on the function set - for those still not using the SandMath on a daily basis ;-)

Maybe I already said it some time ago when this display mode has been discussed here, but I still think that FIXALL is a misleading name for this command: the basic feature of this mode is that the number of digits is not fixed at all. I'd suggest something like FLOAT. Or maybe UNFIX. ;-)

BTW I would love to have the Y-related tests also in the 34s. Testing a condition while leaving X unchanged is a very handy feature in many situations.

Dieter

[Ángel Martin]

(05-02-2015 06:15 AM)Dieter Wrote:  Ah, great – so now there's RCL arithmetics for the 41, at last. As well as a number of new tests. Here the QRG says that ?X> is a "greater than X" test, or ?0< is a "less than zero" test. However, the command name says exactly the opposite: if 0<X then X is greater than zero, not less. This would mean that for instance ?X< 01 actually tests if X > R01. ?!?

You're absolutely correct - I need to double check what's done in the code and then change the text to reconcile function name and descriptions. Thanks for keeping me straight.

(05-02-2015 06:15 AM)Dieter Wrote:  Maybe I already said it some time ago when this display mode has been discussed here, but I still think that FIXALL is a misleading name for this command: the basic feature of this mode is that the number of digits is not fixed at all. I'd suggest something like FLOAT. Or maybe UNFIX. ;-)

I couldn't agree more with you - yet even the humble HP-32S call this FIX ALL so I thought I'd go with the float - I mean with the flow ;-) So the jury is still not done on this one, watch this space for the next rev.

(05-02-2015 06:15 AM)Dieter Wrote:  BTW I would love to have the Y-related tests also in the 34s. Testing a condition while leaving X unchanged is a very handy feature in many situations.

Yes, one of those that kept poping up in my programs was Y<Z? so I decided a no-compromises approach was needed and included the 5 most-frequently used stack regs, {X, Y, Z, T, L} so you can stretch things to the point testing for L<T? if so desired.

[Dieter]

(05-02-2015 07:14 AM)Ángel Martin Wrote:  You're absolutely correct - I need to double check what's done in the code and then change the text to reconcile function name and descriptions. Thanks for keeping me straight.

So let's hope the implementation is right and only the documentation needs an update.

(05-02-2015 07:14 AM)Ángel Martin Wrote:  

(05-02-2015 06:15 AM)Dieter Wrote:  ...I still think that FIXALL is a misleading name for this command: the basic feature of this mode is that the number of digits is not fixed at all.
I'd suggest something like FLOAT. Or maybe UNFIX. ;-)

I couldn't agree more with you - yet even the humble HP-32S call this FIX ALL so I thought I'd go with the float - I mean with the flow ;-) So the jury is still not done on this one, watch this space for the next rev.

The jury might note that there is no FIX ALL command in the 32s (or 33s or 35s or...). There is FIX n and there is ALL, which essentially does what you call FIXALL. So there is no flow to go with – or let's say the flow is "ALL". ;-)

Quote from the 32s II manual:

ALL Format ({ALL})
ALL format displays a number as precisely as possible (12 digits maximum).

Quote from the 35s manual:

ALL Format (ALL)
The ALL format is the default format, displaying numbers with up to 12 digit precision.

BTW I like the way ALL is implemented in the 34s, i.e. with a parameter that sets the threshold where the display switches from, say, 0.033333333 to 3.3333333–02. But this is not trivial, a simple ALL would do, switching to scientific notation for values below 0.01.

Dieter

[Paul Dale]

(05-02-2015 06:15 AM)Dieter Wrote:  BTW I would love to have the Y-related tests also in the 34s. Testing a condition while leaving X unchanged is a very handy feature in many situations.

Space would be the problem here, although I'd love to have the capability.

We've got a lot of conditional tests against X or complex XY. There fourteen tests against 0 and 1, six tests for complex 0, complex 1 and complex i. Two tests for signed zero, seven tests of X against a register and two for complex X against a register pair.

Multiply these by the number of registers for whichs tests should be available (Y, Z, T, L, A, B, C, D, I, J, K, ZT, AB, CD, IL, JK) and then allocate a command table entry for each. Several kilobytes minimum I suspect. The actual code to implement the tests would be relatively small if done properly.


- Pauli

[Ángel Martin]

(05-02-2015 04:16 PM)Dieter Wrote:  So let's hope the implementation is right and only the documentation needs an update.

Yes it was correct. Updated QRG attached.

Still called FIX ALL, and no selector parameter - that one I won't do.

Cheers,
ÁM