Stream: fhirpath
Topic: Date/Time comparison vs equality
Bryn Rhodes (Aug 26 2020 at 14:04):
@Grahame Grieve , starting a new thread for this test:
<test name="testStringQuantityDayLiteralToQuantity" inputfile="patient-example.xml"><expression>'1 day'.toQuantity() = 1 'd'</expression><output type="boolean">true</output></test>
Bryn Rhodes (Aug 26 2020 at 14:05):
This one is actually highlighting an inconsistency in the spec, there is a tracker #28144 to address that, but since UCUM definite-duration quantities and calendar durations above seconds are uncomparable, the result should be empty.
Grahame Grieve (Aug 26 2020 at 19:37):
the issue is units, not math logic. {day} is a non-unit
Bryn Rhodes (Aug 27 2020 at 02:09):
Agree that {day} is a non-unit, this issue is different, specifically day and d are not comparable, so should return empty.
Grahame Grieve (Aug 27 2020 at 02:28):
how are they not comparable?
Bryn Rhodes (Sep 02 2020 at 18:38):
For the same reason that year and 'a' are not comparable, one is a calendar duration, the other is a definite-duration UCUM unit.
Grahame Grieve (Sep 02 2020 at 20:35):
I understand why year and a are not comparable. And month. But I don't understand why for periods of week or less
Bryn Rhodes (Sep 02 2020 at 20:38):
Because of the possibility of leap seconds, only seconds and below are technically exactly the same between the calendar and UCUM units.
Grahame Grieve (Sep 02 2020 at 20:47):
that's not exactly a real world issue.
Bryn Rhodes (Sep 02 2020 at 21:13):
That's fair, but that was the resolution. If the reference implementations behave differently, we can deal with that as an errata I think.
Bryn Rhodes (Sep 04 2020 at 16:32):
Added a comment to this effect to this tracker: https://jira.hl7.org/browse/FHIR-28144?filter=12608
Bryn Rhodes (Sep 04 2020 at 16:40):
And committed a change to the fhir-path-tests to address it. Specifically:
'1 day'.toQuantity() = 1 'd' // true
'1 \'wk\''.toQuantity() = 1 week // true
'1 \'mo\''.toQuantity() = 1 month // empty
'1 \'a\''.toQuantity() = 1 year // empty
Paul Lynch (Sep 30 2020 at 17:02):
Bryn Rhodes said:
And committed a change to the fhir-path-tests to address it. Specifically:
'1 day'.toQuantity() = 1 'd' // true '1 \'wk\''.toQuantity() = 1 week // true '1 \'mo\''.toQuantity() = 1 month // empty '1 \'a\''.toQuantity() = 1 year // empty
@Bryn Rhodes - Your proposed resolution to J#28144 says "above days" should return empty, not "above weeks". (I am just objecting to apparent inconsistency, not to "weeks".)
Bryn Rhodes (Sep 30 2020 at 20:12):
Agreed, and when I applied this to CQL as part of alignment/reconciliation there, I noted weeks everywhere. See Equal for example.
Bryn Rhodes (Sep 30 2020 at 20:12):
I've added this as a note to the comment on that tracker.
Last updated: Apr 12 2022 at 19:14 UTC
