Michael Kay
Michael Kay
Note that the proposal could be considerably simplified if we were to adopt the suggestion in https://github.com/qt4cg/qtspecs/issues/298#issuecomment-1802673000 that arrays should be treated as a subtype of maps.
Yes, doing the examples led me to similar conclusions. But I'm inclined to get something simple working first. It's clearly already quite powerful. That's why I stuck with "replace" for...
Yes, it would probably be a good discipline to start using branches and PRs on the test suite.
I have revised the proposal very substantially. The changes also impact on the semantics of deep lookup, and the concept of labels. An update expression can now have multiple clauses,...
Committed a further revision. This is mainly to address my own concerns rather than comments from others. The new revision reverts to maintaining provenance information for all lookup operators, so...
I've realised that the new proposed syntax `"update" ("map" | "array") [ExprSingle] "{" [UpdateClause] ("," [UpdateClause])* "}"` is problematic. Consider for example `update array array { thing } { replace...
I have dropped the update expression itself from this PR, changing it to focus on lookup with modifiers, returning labeled values. I'll move the update expression itself to a separate...
Fix #1247
I think the changes in this PR were all either abandoned, or reworked as part of other PRs. Therefore closing this one with no further action.
Thanks, yes. I have just spent an hour or so reviewing PR #1283 and I agree it has dependencies on changes that need to be reintroduced or revised.