> Ralph Mellor <ralphdjmel...@gmail.com> wrote: > >> > @r = @r , 'd'; > >> > >> There isn't anything very useful in this behavior though, is there? > > Just to be clear, I wasn't saying I didn't think circular references > should be forbidden, I just specifically meant that you weren't likely > to want the ",=" operator to create them.
Ah. OK. My guess is that when they *are* constructed, they're almost always constructed using `=` and `,`. And that would lead to the same result and display that use of `,=` does. After all, the latter is just shorthand for the former. > But then really, I'm not too likely ",=" for appending additional array > elements, I'm more likely to reach for .push. Indeed. (Or .append.) My tentative strawman proposal based on discussion thus far is: * The doc page showing use of `,=` with a hash might benefit from a parenthetical remark that push/append is the right tool to use with positionals rather than `,=` because, unlike associatives, for which flattening is the only sensible semantics, positionals instead default to *not* flattening (so they more easily retain structure, which is what most folk want most of the time). * Liz's "thoughts welcome" is applicable to all things, and in particular to the gist of self-referential data.