Potential changes to the International RefsetDescriptor records
We found potential issues in the historical descriptor records when implementing new validation - this is a discussion to start socialising the potential updates to fix these issues, and gain feedback on whether or not there would be any impact to end users' implementations if we were to make the changes.
Comments
Couldn't agree more - perhaps we should consider allowing exception listing for the current release cycle only (so that the exceptions expire as soon as the next Release is run). This would slightly increase the maintenance overhead, but the alternative is these kind of issues being obfuscated, which can eventually cause problems.
"this week as International Refset Descriptors were moved into the AU module" - Just to be clear, these entries weren't actually moved into our (RF2) module. (Unless you mean an "Authoring platform module").
My assumption is the whitelisting, is specific to an edition? And these resurfaced because we're new?
I don't think it's technically critical, in that AFAIK the number of implementers using the refset descriptor is close to zero. And even fewer relying on descriptor to be a genuine refinement of the parent refset desciptor.
From my experience people just want to know is it a component (foreign key) or string/integer to configure table schema.
Yes, the moduleId of those refset members switched to be the AU module and at that point, they were no longer recognised as matching the existing whitelist pattern.
Agree with Dion's "vanishingly small" number of people actively using refset descriptors. It may actually be zero. I do of course want to get them right though.
We (SI) should hold some annual review of our whitelist, generate some tickets out of the back of it or something. Certainly I'd like to see a paper trail (via the ticket reference) on any item that we choose to permanently ignore.
CC @Andrew Atkinson @Chris Swires
We see here the danger of exception listing - out of sight, out of mind. This issue has come back to bite this week as International Refset Descriptors were moved into the AU module and therefore showed up because they no longer qualified for the existing whitelisting.
The 'paper' trail for this discussion is: FRI-429, MSSP-1745, INFRA-9188 and then the AU issue came to me today in MSSP-3226.
This is going to take a little time to unpick and we're trying to finish up a sprint cycle this week, so watch this space...
CC @Andrew Atkinson @Matt Cordell @Michael Chu @Yongsheng Gao