dlachieze wrote: ↑Fri Sep 28, 2018 8:23 am

The first HP calculator with 2 variables stat functions is the HP-55. On the HP-55 MEAN and SDEV are overwriting X and Y, no stack lift. This is clearly documented in the manual.

This is also how it works in the following 67, 27, 29C/19C, 41C and 42S. So this is the original stack handling being in use from 1975 to 1988.<br/>

It appears this has been changed for the Spice and the Voyager follow-ups.

For backward compatibility the 41C needed to work the same way as the 67. Same for the 42S.

The first HP pocket calc "with statistics" was the HP-45, summing and evaluating one variable only. It was followed by the HP-25/25C. Thereafter, all pocket calcs sported 2D statistics.

The first HP featuring 2D statistics and L.R. as a built in function was the HP-55. It was followed by the HP-22, 27, 29C/19C, and 67/97 (though the latter two pairs don't feature L.R.). All these overwrite x and y when statistical functions returning 2 values are called.

Thereafter, the HP-32E, 33E/C, 34C, 10C, 11C, 12C, and 15C all featured 2D statistic and built-in L.R. When statistical functions returning 2 values are called, all these models push the output on the stack. I can't tell about the way the HP-37E and 38E/C handle the stack but assume it's the same.

The HP-41C and HP-42S returned to the old paradigm overwriting two stack levels.

And further RPN calcs didn't return two statistical values anymore.

Personally, I think the Spice and Voyager way of dealing with double returns is more logical for an RPN calc. Hence, we implemented it in the WP-34S and 31S. And we are going to implement it in the 43S as well.

DM42 SN: 00041 --- Follower of Platon.

HP-35, HP-45, ..., HP-50, WP 34S, WP 31S, DM16L