Nicholas Clark wrote:


So I assume, that the returned PerlUndef should be put into the
aggregate, if there was none before access.


I believe yes, it must, for consistency. Is there an op to copy the value out of an aggregate, without changing the aggregate? The (shallow copy) value assignment, needed to implement

I think, we need another kind of keyed opcodes that do assign


assign P0, P1[key] # assign in PMC, in PMC, in (INT)?KEY

Same for other LHS types. The LHS has to exist before, and get the value of the aggregate member, which doesn't get autovivified like set should do. Its the same as in the "set vs assign" thread.


Nicholas Clark

leo







Reply via email to