Stream: IG creation
Topic: feature request
Jens Villadsen (May 28 2019 at 12:11):
Could the version of the IG tool be added to the qa page?
Grahame Grieve (May 28 2019 at 13:00):
ok in the next version (not the one I am releasing right now)
Grahame Grieve (May 29 2019 at 00:04):
on the subject of feature requests... would IG publishers opt in to collecting extension usage statistics to help with specification quality issues?
David Pyke (May 29 2019 at 00:05):
I would. What do I need to do?
Grahame Grieve (May 29 2019 at 00:05):
say yes somewhere
David Pyke (May 29 2019 at 00:05):
Yes somewhere
Grahame Grieve (May 29 2019 at 00:09):
what about opt-out...?
David Pyke (May 29 2019 at 00:12):
I can't see why we wouldn't keep stats on usage. That would make it far easier to know when an extension should be made part of the core spec.
Grahame Grieve (May 29 2019 at 00:14):
right that's the question
David Pyke (May 29 2019 at 00:15):
So, do we make it a flag in the IG resource or are you going to make it some kind of registry entry?
Grahame Grieve (May 29 2019 at 00:20):
don't think it should be in the resource. Either in the json file or somewhere else
David Pyke (May 29 2019 at 00:20):
An entry in package-list.json makes good sense
Grahame Grieve (May 29 2019 at 00:20):
why there?
David Pyke (May 29 2019 at 00:21):
Makes it possible to differentiate between versions. If someone is no longer interested in participating for an old version, they can specify which they do want
Grahame Grieve (May 29 2019 at 00:22):
but I only build the current version....
David Pyke (May 29 2019 at 00:24):
Right, but tracking usage could be for any version. We don't need to only track what's used in the current, especially when it's usable for regions/implementations/etc. that are using past versions. If suddenly a DSTU2 extension is suddenly all the rage, then it can be looked at for inclusion into current
David Pyke (May 29 2019 at 00:25):
I could be incredibly wrong about this
Grahame Grieve (May 29 2019 at 00:42):
I was referring to the version of the IG, not the version of FHIR
Grahame Grieve (May 29 2019 at 01:58):
Grahame Grieve (May 29 2019 at 01:58):
something like that
Lloyd McKenzie (May 29 2019 at 02:01):
Couldn't the IG publisher detect usage and spit it out somewhere? Publishers shouldn't have to do anything. Could even differentiate whether extension was 'mustSupport' or not.
Grahame Grieve (May 29 2019 at 02:02):
well, we want Telus etc to actually choose to be part of it. Otherwise, yes, spit it out and submit it somewhere
Lloyd McKenzie (May 29 2019 at 02:05):
So a flag you choose to invoke the publisher with? Hl7 frameworks will do so by default, others have to choose. One challenge is that there's no way to differentiate IGs with zero current (and potentially future) implementations from those with thousands.
Grahame Grieve (May 29 2019 at 02:06):
of course that's a limitation in the whole approach, yes. but some data is better than no data.
Lloyd McKenzie (May 29 2019 at 02:06):
Agree
Jens Villadsen (May 29 2019 at 10:42):
yep
Grahame Grieve (May 29 2019 at 12:23):
ok: http://test.fhir.org/usage-stats
Grahame Grieve (May 29 2019 at 13:13):
to opt-out, using json:
"usage-stats-opt-out" : true,
to opt out using the new ig only approach, add this to the ini file:
[IG]
usage-stats-opt-out=true
Grahame Grieve (May 29 2019 at 13:14):
you can't opt out if you HL7 or you are using the ci-build
Grahame Grieve (May 30 2019 at 21:18):
for those who are interested, @David Hay is experimenting with views of this
Grahame Grieve (May 30 2019 at 21:18):
v
Grahame Grieve (May 30 2019 at 21:18):
http://clinfhir.com/igAnalysis.html
David Hay (May 30 2019 at 22:41):
So there's a (very) basic visualizer up there. There's a lot you could do with this data (and I suspect it will get very big very quickly), so would be good to understand what people would like to get out of it. Simple views on the raw data might not cut it for long...
David Hay (May 30 2019 at 22:42):
Oh and currently this works against a snapshot of the data - hope to have a real time view soon...
Grahame Grieve (May 30 2019 at 22:43):
it's already useful
David Hay (May 30 2019 at 23:04):
good-oh...
David Hay (May 31 2019 at 00:10):
Added a view of an IG (the extensions used by path is kinda cool)
David Hay (May 31 2019 at 00:14):
What would be nice would be to pull down details of the extensions - perhaps from the simplifier registry?
Grahame Grieve (May 31 2019 at 00:19):
what details? And you are not guaranteed to get the extensions - stats may come from non-public IGs
David Hay (May 31 2019 at 00:28):
dataType for example, description. But may be going beyond the purpose of the view...
Grahame Grieve (May 31 2019 at 00:28):
I might add datatype
Grahame Grieve (May 31 2019 at 00:49):
I have. it will gradually get populated
David Hay (May 31 2019 at 03:28):
Good-oh. Let me know when the CORS fix is uploaded...
Grahame Grieve (May 31 2019 at 03:44):
done
David Hay (May 31 2019 at 06:02):
and update pushed...
Sarah Gaunt (May 31 2019 at 23:32):
@David Hay Looks great - can see this coming in super handy...
David Hay (May 31 2019 at 23:52):
Thanks Sarah - let me know if you can think of any improvements (or issues!)
Grahame Grieve (May 31 2019 at 23:54):
one for me - the struture definition extension count, each combination of path | extension | IG is counted separately on the resource view. I think that it should be just once for the extension, with paths/IGs as a detail view
David Hay (Jun 01 2019 at 02:50):
This is the 'view by resource type' - right? The count (in the badge to the right of the type) should be the unique number of extensions, regardless of the number of IG's or paths where it is used. So for AllergyIntolerance it should be 4 (at the moment)
Grahame Grieve (Jun 01 2019 at 02:54):
yes
David Hay (Jun 01 2019 at 03:35):
ok
David Hay (Jun 01 2019 at 03:44):
and done...
Jens Villadsen (Nov 27 2019 at 14:59):
@Grahame Grieve For the page that renders the capabilitystatement table, could that also include the search/include/revInclude parameters?
Grahame Grieve (Nov 27 2019 at 15:46):
it could. I'd like to see what other people think about this?
John Moehrke (Nov 27 2019 at 16:07):
yes, please
Lynn Laakso (Nov 27 2019 at 17:01):
(deleted)
David Pyke (Nov 27 2019 at 17:36):
yes, please
Eric Haas (Nov 27 2019 at 19:22):
like this?
Grahame Grieve (Nov 27 2019 at 19:42):
Too much information?!
Jens Villadsen (Nov 27 2019 at 21:07):
thats fine for me as a first shot. Separate page or just presented below existing table?
Last updated: Apr 12 2022 at 19:14 UTC
