sys, phydat: omit scale to string conversion if not applicable

[smlng]

based on #7193

[kaspar030]

based on #7193

please state:

• what does this PR do
• how does it do it

[smlng]

This PR adapts scale to string conversion to work with multi dimensional units like square (m^2) and cubic meters (m^3). For further info see discussion in #7193

[miri64]

#7193 was merged, please rebase.

[smlng]

rebased

[smlng]

another option would be to leave scale_to_str as is and make the check and rescaling for certain units in the calling function phydat_dump.

[haukepetersen]

partly NACK: IMHO we can simplify to the following:

char phydat_scale_to_str(int8_t scale, uint8_t unit)
{
    switch (unit) {
        case UNIT_M2:
        case UNIT_M3:
        case UNIT_TEMP_C:
        /* ... add all others for which scale names don't make sense */
        return '\0';
    }
    
    switch (scale) {
...

Also, the calls to phydat_str need to be adapted...

[miri64]

another option would be to leave scale_to_str as is and make the check and rescaling for certain units in the calling function phydat_dump.

Yes, alternatively this check can be done in phydat_dump(). What do you think @haukepetersen?

[smlng]

would be less intrusive, too - I guess

[smlng]

I changed the handling as suggested

[miri64]

I changed the handling as suggested

Now the PR needs a new title ;-)

[smlng]

depends: which version do you and @haukepetersen like better?

[miri64]

Afaik Kelvin (a SI unit) my be used with SI prefix.

[miri64]

(see e.g. https://en.wikipedia.org/wiki/Orders_of_magnitude_(temperature))

[smlng]

though completely unrealistic for RIOT, but fine - I'll remove it

[miri64]

Well it's one check less saving a few byte of ROM ;-)

[haukepetersen]

I don't like to put it into phydat_dump, as the phydat_scale_to_str() is/will be used also in other contexts. So I really prefer the 'white-listing' approach for simply returning \0 for a number of units.

[smlng]

@haukepetersen currently phydrat_scale_to_str() simply does what it says, without any dependencies, which I think is good. The additional check for certain units would require to pass the unit to that function, which I think would change its purpose.

IMHO for now its fine to have it in phydrat_dump, we may at an intermediate function later on - who is using scale_to_str, too?

[haukepetersen]

ok, fine with me. Then lets keep it like this.

[haukepetersen]

please squash

[smlng]

done

[haukepetersen]

Could you quickly fix the whitespace issues murdock is complaining about? Just squash the fixes right in...

[smlng]

whitespace issues

damn, my bad - anyway, fix amended.

[haukepetersen]

Murodck is still unhappy, but the cause could be unrelated to your changes. Would you fix it anyway? Thx

[smlng]

I think I made non-commit, with[out] actually pushing the fix - let's see what Murdock says now

[haukepetersen]

All green -> lets go