Re: [patch] for PR 18040

[kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)]

    Your opinion on my alternative suggestion - hoist the type conversion
    to the outermost type, thus reducing the nested case to the unnested
    case - would be appreciated (and please read the whole message before
    answering the question, because I address another objection below).

Aside from the issue of possible quadratic behavior, I don't completely
understand the proposal here, so I can't fully comment on it.  I should
point out, though, that some of these expressions in practice are
quite complex: there are lots of implicit conversions and dereferences
generated by the front end.

    I didn't say this had to be captured in the type-based alias sets, I
    said it had to be comprehensible to the alias analyzer.  Those two
    assertions are not the same thing.

No, but your comment about the GNAT alias set computation was sounding like
you were talking about type-based alias sets.

    > This stuff is *very* tricky and, as I said, we've been throught it
    > before.

    The PR indicates that the solution which has already been implemented
    does not work.  Thus, the entire topic is open for reexamination.

The PR indicates that a particular optimizer has problems with the
overall approach.  One way of dealing with that is to change the approach.
But a more local way is to fix that particular optimizer.