Ruling out arbitrary cost copy construction?

[dsimcha]

== Quote from Andrei Alexandrescu (SeeWebsiteForEmail at erdani.org)'s article
> The GC shouldn't be able to destroy things naively concurrently with
> other threads. Currently all threads are frozen; I'm not sure whether a
> frozen thread commits its writes, so that needs to be verified.
> Andrei

How could freezing a thread not commit all its writes?  If it didn't, then pointer
updates wouldn't be committed and things might get GC'd when they shouldn't.