grouper-subject-identifier

Beth Halsema Today at 08:07
Chris or Chad, the subjectID references in the documentation and examples have referred to the system of record's unique identifier.  Our university has such a unique identifier.  In general, the unique identifier is correctly assigned.
But it is not guaranteed to be constant.  Was the ID erroneously associated with another?  Or maybe it wasn't and it should?  We call them splits or merges.   And as a result, it is then changed.   It really isn't opaque, because others are able to manipulate it.
Would you recommend having one that is truly internal to the InCommon Trusted Access Platform ecosystem?  One that only it knows about, and it doesn't change.  The subject source would have this truly opaque identifier that only "we" know about it which maps to the system of record ID, username, name, etc.etc.etc.
Other than it is yet one more unique identifier and that the processing of the nightly data must incorporate it into the data, can you think of a reason not to do it?
Thank you! (edited)
6 replies

Chris Hyzer  12 hours ago
opaque means its not based on name generally, and wont change when someone changes their name.  if its based on initials, but you have a policy not to change it, that is ok.  You can adjust the subjectId in Grouper if things change.  Of course you will have splits and merges, but it should happen very infrequently and only because matching isnt 100% right?  whats an example of the id?  if everyone has in the institution, i think you should probably use it, but we can discuss more :slightly_smiling_face:

Chris Hyzer  12 hours ago
at penn users who can log in have a pennkey which for me is mchyzer.  we also all have a pennid which is the IDM id.  mine is 10021368.  The pennkey can change, so that is not our subjectid.  the pennid doesnt change but sometime we have dupes and need to merge.  the pennid is our subjectId

Chris Hyzer  12 hours ago
10021368 is not based on name, so it is considered opaque

Beth Halsema  11 hours ago
Purdue University's unique ID is a ten-digit number that is not based on the name.  It can change as the result of a demographic correct/incorrect matching.

Chris Hyzer  11 hours ago
i think that sounds right for subjectid :slightly_smiling_face:

Chris Hyzer  11 hours ago
if you have a process for merging, you can call the member change subject web service, or it could be a manual process to see what grouper has and delete one or call member change subject via gsh.  At penn we generally just ignore this case and let it shake out (new pennid gets all the basis stuff automatically, the only pennid just goes away).  But if someone had ad hoc memberships the member change subject call will fix that
https://spaces.at.internet2.edu/display/Grouper/Member+change+subject
	Beth Halsema Today at 08:07
	Chris or Chad, the subjectID references in the documentation and examples have referred to the system of record's unique identifier. Our university has such a unique identifier. In general, the unique identifier is correctly assigned.
	But it is not guaranteed to be constant. Was the ID erroneously associated with another? Or maybe it wasn't and it should? We call them splits or merges. And as a result, it is then changed. It really isn't opaque, because others are able to manipulate it.
	Would you recommend having one that is truly internal to the InCommon Trusted Access Platform ecosystem? One that only it knows about, and it doesn't change. The subject source would have this truly opaque identifier that only "we" know about it which maps to the system of record ID, username, name, etc.etc.etc.
	Other than it is yet one more unique identifier and that the processing of the nightly data must incorporate it into the data, can you think of a reason not to do it?
	Thank you! (edited)
	6 replies

	Chris Hyzer 12 hours ago
	opaque means its not based on name generally, and wont change when someone changes their name. if its based on initials, but you have a policy not to change it, that is ok. You can adjust the subjectId in Grouper if things change. Of course you will have splits and merges, but it should happen very infrequently and only because matching isnt 100% right? whats an example of the id? if everyone has in the institution, i think you should probably use it, but we can discuss more :slightly_smiling_face:

	Chris Hyzer 12 hours ago
	at penn users who can log in have a pennkey which for me is mchyzer. we also all have a pennid which is the IDM id. mine is 10021368. The pennkey can change, so that is not our subjectid. the pennid doesnt change but sometime we have dupes and need to merge. the pennid is our subjectId

	Chris Hyzer 12 hours ago
	10021368 is not based on name, so it is considered opaque

	Beth Halsema 11 hours ago
	Purdue University's unique ID is a ten-digit number that is not based on the name. It can change as the result of a demographic correct/incorrect matching.

	Chris Hyzer 11 hours ago
	i think that sounds right for subjectid :slightly_smiling_face:

	Chris Hyzer 11 hours ago
	if you have a process for merging, you can call the member change subject web service, or it could be a manual process to see what grouper has and delete one or call member change subject via gsh. At penn we generally just ignore this case and let it shake out (new pennid gets all the basis stuff automatically, the only pennid just goes away). But if someone had ad hoc memberships the member change subject call will fix that
	https://spaces.at.internet2.edu/display/Grouper/Member+change+subject