Dan Sugalski <[EMAIL PROTECTED]> wrote: > The question isn't whether they *do*--rather it's whether they > *should*. The current implementation's not all that relevant except > as a demonstration of one set of semantics. What I want to do is work > out what the 'proper' semantics ultimately should be.
I think that the static "inheritance" inside classes/ is ok. We have to rearrange classes hierarchy, though. The special MMD based Integer and Float classes are no more special, array classes still need finishing. A nice to have would be a more sophisticated MI scheme. It's ok inside the PMC files, e.g. from orderedhash.pmc: void mark () { SUPER(); PerlHash.SUPER(); } but the build dependencies are created manually. leo