Difference between revisions of "Procedures for Datatypes R2 changes and associated RIM changes"
m |
|||
Line 16: | Line 16: | ||
Would implementors be confused by this? | Would implementors be confused by this? | ||
− | Lloyd: Adding it to the RIM doesn't break anything. If committees choose to pre-adopt, then they can't become normative until the datatypes become normative. | + | Lloyd: Adding it to the RIM doesn't break anything. If committees choose to pre-adopt, then they can't become normative until the datatypes become normative. I think that's a reasonable path |
Grahame: I'm not so sure about that last bit: if a committee has a model that says an attribute has type X, | Grahame: I'm not so sure about that last bit: if a committee has a model that says an attribute has type X, |
Revision as of 15:39, 30 April 2007
Introduction
This issue comes out of a Datatypes R2 proposal. There is a proposal to create a new datatype, and change some RIM attributes to use the new attribute. The proposal is Datatypes R2 Issue 57
The problem is when the change should be made
Problem
The datatype doesn't really exist until datatypes R2 passes normative ballot. Can we pre-adopt the datatype in the RIM, like we did for R1? If we do this, do we start issuing a schema that preadopts R2 changes? Would implementors be confused by this?
Lloyd: Adding it to the RIM doesn't break anything. If committees choose to pre-adopt, then they can't become normative until the datatypes become normative. I think that's a reasonable path
Grahame: I'm not so sure about that last bit: if a committee has a model that says an attribute has type X, and the RIM changes to type Y, in which X is not a valid substitution for Y, does the committee have any choice but to change with the RIM?