"300 degC" wrongly converted to "300 K" by toSI()

[jkotan]

The problem is related to #2499 but it was not fixed by the corresponding PR.

When we convert units from degC to K in SciCat they are wrongly converted by toSI() i.e. without adding the offset (SciCatProject/scicat-backend-next#926 ).

In physics you never use Celsius to describe a difference of two various temperatures by always Kelvin
so during conversion from degC to K the offset should be always added.

You can workaround the problem using twice unit() , to() and toSi() but it looks complicated.

Is this behavour expected?

To Reproduce

>  const { unit, version } = await import("mathjs");

> version
'12.1.0'

>  unit(300. "degC").toSI().toJSON()
{ mathjs: 'Unit', value: 300, unit: 'K', fixPrefix: true }

>  unit(300, "degC").to(unit("degC").toSI()).toJSON()
{ mathjs: 'Unit', value: 573.15, unit: 'K', fixPrefix: true }

[josdejong]

Thanks for reporting. At first sight this indeed looks like a bug in Unit.toSI to me, this method does not contain any logic to recon with a temperature offset like Unit.to does.

Any thoughts on this @costerwi or @gwhitney?

[gwhitney]

I mean the whole situation is super complicated by whether a given units quantity is a specific temperature, or a difference in temperatures. I can't keep straight in my head the convention we were heading to between what units to use for which; I think maybe it was that degC should only be used for specific temperatures, and K for either specific temperature or differences? If so, then yes, using the offset when converting should happen. So sounds like this is a bug. The real best way to handle these things would be to get the modularization with respect to a separate Unit package done, and then put the energy into fixing that Unit package... sorry not to be of more help.

[josdejong]

Thanks for your inputs Glen.

I think the issue here is that the behavior of toSI() and to() are not consistent:

math.unit("300 degC").toSI().toString()   // "300 K"
math.unit("300 degC").to("K").toString()  // "573.15 K" 

Maybe .toSI() is just not on par with .to(). That could be related to the changes in .to() that we made in #2501.

[costerwi]

Note this result is consistent between the two methods:

math.unit("300 degC^-1").to("K^-1").toString()  // "300 K^-1"
math.unit("300 degC^-1").toSI().toString()   // "300 K^-1"

I prefer the behavior of the existing .to() method and I think the toSI() example in the subject of this issue looks like a bug. Maybe it can be rewritten to identify the correct SI unit and then convert to it using .to().

[josdejong]

Maybe it can be rewritten to identify the correct SI unit and then convert to it using .to().

Reusing that logic would be nice indeed.

[josdejong]

This issue has been fixed now in v12.3.0 via #3118.