Fix a typo in the quantity tests

[jingtang10]

My guess is that intention of the first 4 test cases for quantity is to test =, ~, !=, and !~ operators in order. So I corrected testQuantity4 to test !~ instead of using ~.

[jingtang10]

I want to add that different FHIRPath engines implement this differently:

fhirpath-lab

This is surprising to me since the spec states very clearly:

When comparing quantities for equality, the dimensions of each quantity must be the same, but not necessarily the unit ... The comparison will be made using the most granular unit of either input.

[brynrhodes]

Note sure on this one, since the equivalent for decimals says precision is ignored, but if you perform the conversion as specified in Quantity, the conversion to mg would be performed first, resulting in a comparison of 4000 to 4040, so shouldn't this be not equivalent?

https://hl7.org/fhirpath/#equivalent

[jingtang10]

Bryn and I discussed this one and we think the spec is not clear.

In the quantity equivalence section, there's this sentence:

The comparison will be made using the most granular unit of either input.

So that is to indicate the quantities in question should be converted to mg before comparison.

Then, how is equivalence actually determined? There is no mention in the spec, it should say that the the numeric parts of the quantities should be equivalent following the decimal equivalence definition - which includes this sentence: comparison is done on values rounded to the precision of the least precise operand.

Now there is a question whether the precision is preserved after converting 4g to 4000mg. So although the value here is 4000mg, since the precision is only the first significant digit (actually i don't know if this is valid since it's not expressed in scientific notation, but let's assume this is expressed in scientific notation.) comparing it with 4040 mg still would result in equivalence.

But if this is true, this sentence: The comparison will be made using the most granular unit of either input. would never be useful?... since we might as well just use the least granular unit of the inputs?

Moreover, Bryn raised his concern that this is actually stepping on the toes of UCUM.

[grahamegrieve]

ok

[grahamegrieve]

Is there a jira task for the committee to clarify this?

[brianpos]

Yes there are several open issues for it.
This is the most direct exat issue as logged.
https://jira.hl7.org/browse/FHIR-53249
But also https://jira.hl7.org/browse/FHIR-49525 and https://jira.hl7.org/browse/FHIR-53250