FHIR Chat · restricting value[x] without slicing

Stream: shorthand

Topic: restricting value[x] without slicing

Patrick Werner (Oct 13 2020 at 15:37):

I just tried to fix value[x] to valueCodeableConcept (inside of a component)

Patrick Werner (Oct 13 2020 at 15:38):

I used: * component[dna-chg].value[x] only CodeableConcept

Patrick Werner (Oct 13 2020 at 15:39):

which creates a type based slicing, which we don't want. Is there an alternative syntax to just fix the type of value[x] ?

Chris Moesel (Oct 13 2020 at 15:58):

Hi @Patrick Werner. If I recall correctly, the rule you noted above (an only rule) should not create a type slice. What typically creates a type slice is if you access value[x] using the type-based path, e.g. * component[dna-chg].valueCodeableConcept from MyValueSet. Are you doing that? If so, you can change those rules to use value[x] instead of valueCodeableConcept -- and that should avoid the type slicing.

Chris Moesel (Oct 13 2020 at 16:00):

BTW -- we're considering changing how SUSHI exports this so that even if you say valueCodeableConcept, SUSHI would not use a type slice if value[x] has already been narrowed to only CodeableConcept anyway. This is something that would change current output (for current IGs), however, so we're considering it carefully. Until then, avoiding the type-specific property names in the paths should do the trick.

Patrick Werner (Oct 14 2020 at 10:15):

it worked! Thanks @Chris Moesel !!

Patrick Werner (Oct 14 2020 at 10:16):

Chris Moesel said:

BTW -- we're considering changing how SUSHI exports this so that even if you say valueCodeableConcept, SUSHI would not use a type slice if value[x] has already been narrowed to only CodeableConcept anyway. This is something that would change current output (for current IGs), however, so we're considering it carefully. Until then, avoiding the type-specific property names in the paths should do the trick.

I think that would be a great change, seems to be much more intuitive (at least to me)

Patrick Werner (Oct 14 2020 at 10:26):

btw: if i do this * valueCodeableConcept from http://loinc.org/vs/LL1971-2 (required) (on Observation.value) doesn't introduce a slice.
I expected that i have to change this also to: * value[x] from http://loinc.org/vs/LL1971-2 (required)

Chris Moesel (Oct 14 2020 at 12:11):

Patrick Werner said: