Re: Garbage Collection in Ada

[dewar@merv.cs.nyu.edu (Robert Dewar)]

Lars says

"
With standard sanction the potential problems you and others (optimizer
tricks that might trip off the collector) would also be under control.
If there is GC it is controlled by the same people that control the
optimizer. With sanctioned GC there would be restrictions on what
optimizer tricks would be allowed.
"


No, you are confused, if a language expects GC, and if a proper GC is
incorporated, there is no respect in which the presence of GC inhibits
optimizations. Where did you get this idea?

It is true that a conservative collector depends on assuming the absence of
no optimization, but there is no way that a language definition could
somehow "sanction" a conservative collector.

At best it could say something like "An implementation is free to remove
blocks  from memory that cannot be further referenced", but that does not
say no use of virtual origins.

You can't have something in a language standard that says "the implementation
must work correctly with a conservative collector that works as follows" !

At least I *trust* that no one would suggest such a statement in a language
standard, but really I know better, people suggest all sorts of incredible
things.