FHIR Chat · ValueSet.expansion.parameter

Stream: terminology

Topic: ValueSet.expansion.parameter

Jack Bowie (Aug 01 2018 at 14:50):

Pardon the rather basic question, but in R4 what is returned in the expansion.parameter elements? The documentation says they are "all the parameters that affect the contents of the expansion". This is a pretty broad statement. So are these the attributes that are named in the VS expression or are they duplicates of the $expand In Parameters like filter and activeOnly? If the latter, do only those parameters with non-default values need be specified? The example does not help. Thanks.

Grahame Grieve (Aug 01 2018 at 20:00):

not things from the VS expression

Grahame Grieve (Aug 01 2018 at 20:01):

and only parameters with nondefault values

Rob Hausam (Aug 01 2018 at 22:07):

the tracker and resolution for GF#15144 states that it should include "all of the parameters, including server defaults"

Grahame Grieve (Aug 01 2018 at 22:08):

we should clarify this then - server defaults that are not default values

Rob Hausam (Aug 01 2018 at 22:10):

yes - it isn't clear to me exactly what is meant by "server defaults that are not default values"

Grahame Grieve (Aug 01 2018 at 22:20):

well, the default value for activeOnly is false. So we don't echo this in the expansion parameters if the client doesn't specify it, or specifies it as false, or if the server default value is false.

Grahame Grieve (Aug 01 2018 at 22:20):

but if the server default is true then that should go in the expansion parameters

Peter Jordan (Aug 01 2018 at 23:17):

I need to implement that! BTW should the name of the expansion parameter containing the version of the Value Set be 'valueSetVersion' (mirroring the name of the input parameter) or simply 'version'?

Jack Bowie (Aug 02 2018 at 00:03):

And count and offset I presume?

Grahame Grieve (Aug 02 2018 at 00:11):

yes to count and offset. @Peter Jordan there's special parameters defined in R4 for version

Peter Jordan (Aug 02 2018 at 00:34):

Thanks - I can only see special parameters relating to the version of a Code System, but I guess the Value Set version would be superfluous as that's contained elsewhere. BTW: what's the difference between expansion.offset and the offset expansion.parameter?

Grahame Grieve (Aug 02 2018 at 00:44):

ah yes I was thinking of the code system versions. the value set version - yes - elsewhere

Grahame Grieve (Aug 02 2018 at 00:45):

what's the difference between expansion.offset and the offset expansion.parameter?

oops. have we (I) documented both?

Rob Hausam (Aug 02 2018 at 00:50):

I get the point about "So we don't echo this in the expansion parameters if the client doesn't specify it, or specifies it as false, or if the server default value is false" - but wouldn't it be simpler and clearer if we just explicitly included the values of the parameters regardless of whether they are the default value or not?

Grahame Grieve (Aug 02 2018 at 01:06):

most implementers would regard that as needless verbosity

Rob Hausam (Aug 02 2018 at 01:07):

possibly - I'm fine with not doing that if there's general agreement

Jack Bowie (Aug 02 2018 at 01:40):

Including only non-defaults works for me.

Peter Jordan (Aug 02 2018 at 03:37):

what's the difference between expansion.offset and the offset expansion.parameter?

oops. have we (I) documented both?

ValueSet.expansion.offset..."If paging is being used, the offset at which this resource starts. I.e. this resource is a partial view into the expansion. If paging is not being used, this element SHALL NOT be present."

ValueSet.expansion.parameter["offset"]..."Paging support - where to start if a subset is desired (default = 0). Offset is number of records (not number of pages)" - this echoes the description of the offset input parameter to an $expand request.

Grahame Grieve (Aug 02 2018 at 05:00):

sigh better fix that

Jack Bowie (Aug 02 2018 at 15:58):

Should a DateTime data type be supported in ValueSet.expansion.parameter to accommodate a 'date' parameter in the expansion request, or is the ValueSet version in the result sufficient?