: fvalue [inlined] 2value ;

[robzed]

When I'm compile-to-flash, fvalue's don't appear to have the correct value. And resetting doesn't show the correct value?

compile-to-flash
#include float_compatibility.fth
#include config.fth

Files can be found here: https://github.com/robzed/ukmarsbot_forth/tree/master/for_rp2040

[tabemann]

I tried reproducing this and was unable to. However, I should warn you that many fixed point numbers with digits to the right of the decimal point will not look "correct" due to not having any exact binary value.

[robzed]

As you know, getting issue reports is half figuring what the tester/user means and half where the problem is! Fixing often is often easy after these two.

I think this is a big gotchya. If we can't fix this somehow in code then we should mention this next to the value and 2value descriptions and on the wiki in the section about Value.

[robzed]

By the way I found value/2value ideal for the use case here. They are effectively modifiable constants - which really works well for this configuration variables.

True - extra work would be required to make modified values persistent - but I think this is good. I don't want to save temporary test values every time I try something and lose the original value.

So definitely worth the effort it too you to add values - thanks.

[robzed]

I think I've figured out the problem with config.fth. Even after a reboot the value isn't correct - because they are calculated from other values

I figured this out by adding a print before defining:
calc_deg/mm_diff fdup f. fvalue DEG/mm_DIFFERENCE 0,0000003350 ok

It should be 0,8952465557.

To repeat - of course the value we are storing isn't correct here ... because it's calculated based on other values, which are not correct until reboot. I'll have a think about how to get around this - but this use of value/2value isn't going to work here as it is.

Extra info

Using the following:

compile-to-flash
#include float_compatibility.fth
#include config.fth

For reference below are the results I get.

from end of config.fth

: calc_deg/mm_diff ( -- f ) 180,0 2,0 f/ WHEEL_CIRCUM f/ ; ok
calc_deg/mm_diff fvalue DEG/mm_DIFFERENCE ok
: calc-config ok
calc_wheel_circum to WHEEL_CIRCUM \ mm ok
calc_pulses/rev to PULSES/REV ok
calc_mm/count to BASIC_mm/count ok
calc_mm/count_left to mm/COUNT_LEFT ok
calc_mm/count_right to mm/COUNT_RIGHT ok
calc_deg/mm_diff to DEG/mm_DIFFERENCE ok
; ok
: show-config ok
." mm/COUNT_LEFT " mm/COUNT_LEFT f. cr ok
." mm/COUNT_RIGHT " mm/COUNT_RIGHT f. cr ok
." DEG/mm_DIFFERENCE " DEG/mm_DIFFERENCE f. cr ok
; ok
DEG/mm_DIFFERENCE .s [ 1573941093 1917799667 ] ok
**f. 1917799667,3664617177 ok
reboot
Welcome to zeptoforth
Built for rp2040, version 1.1.0, on Mon Sep 11 07:21:28 PM CDT 2023
zeptoforth comes with ABSOLUTELY NO WARRANTY: for details type `license'
ok
DEG/mm_DIFFERENCE .s [ 1439 0 ] ok
f. 0,0000003350 ok
calc-config ok
DEG/mm_DIFFERENCE .s [ -449912617 0 ] ok
f. 0,8952465557 ok

[tabemann]

What I would probably do is to create the initial values as constants so they can be utilized when compiling to flash and use the generated values as the default values for the values in such a fashion that the values are never read until after rebooting.

[robzed]

That's a good idea using constants as defaults.

[tabemann]

One other note - there is no need to make fconstant, fvariable, or fvalue inlined because in 99% of use cases these will be executed in interpretation mode.

[robzed]

Agreed.

[robzed]

Fixed

[tabemann]

@robzed Shall I close this issue, or is there anything else here? Thanks for bringing this up, BTW, as you probably have noticed I sometimes forget that other people don't know all the ins and outs of zeptoforth.

[robzed]

Fine to close