On Fri, Jul 5, 2019 at 8:21 PM Stefan Monnier <monnier@iro.umontreal.ca> wrote:
> >> His patch can/should be made slightly more efficient by only doing the
> >> Fcopy_sequence on those hash-tables that are in purespace.
> >
> > How do we test for that?
>
> With PURE_P?

That only works before we dump, I think?

> > I'm not sure how difficult it would be, but we could dump the ->index,
> > ->next, ->hash vectors as Qnil (so not include the actual vectors in
> > the dump), which would make the dump slightly smaller and give us a
> > better test than h->count < 0:
>
> Except that it can't be done at the time of purecopy but only at the
> time we do the actual dump because the purecopied hashtable may be used
> in-between (which is also why count is kept positive by purecopy and is
> only made negative later).

Indeed. I'm attaching a proof of concept that we can simply freeze the
hash tables when dumping and thaw them when loading a dump, which
includes rehashing. Do you happen to know why it wasn't done that way?
I quite enjoyed removing all the scattered hash_rehash_if_needed()s.