Dear Denis, group, On Wed, Oct 07, 2020 at 01:13:15PM +0000, [email protected] wrote: > This is the last of the NWIs from 4 years ago > https://www.ripe.net/ripe/mail/archives/db-wg/2016-June/005272.html > > For this one I am not even sure what is being asked for and I don't > agree with some of the points in the problem statement. Maybe Job can > expand on it. Any comments appreciated...
I recall correctly some people argued that adding or removing attributes can potentially result in brittleness on the data consumer side. A formal solution would be to mark each object with some kind of 'object schema version', and a reference to the applicable data model. (Think DTD in XML context, or how things work in YANG). In the abstract this idea sounds great, but I fear there are no takers to do the heavy lifting of speccing it out, and updating the software ecosystem to benefit from schema-awareness. Retrofitting modern day insights into data model handling onto RPSL is perfectly achievable, but a significant project in terms of time. All in all recommending heavy RPSL consumers to subscribe to [email protected], to read the release notes, and test against the RIPE test environment seems a simpler solution. I don't mind abandoning NWI-6. Kind regards, Job
