Re: GUILE 2/3 and string encoding cost

[David Kastrup]

Han-Wen Nienhuys <address@hidden> writes:

> On Fri, Jan 24, 2020 at 10:51 AM David Kastrup <address@hidden> wrote:
>
>>
>> > What do you mean with "heap is collected"?
>>
>> "Collected" is probably the wrong expression.  Sweeped and marked.  The
>> proposed behavior by Guile developers is not to bother with individual
>> mark hooks and just let the whole heap be marked and sweeped.
>>
>
> what is the documented way of disabling the hooks?

Just don't register them, I suppose.

All of our Guile allocation stuff by now is supposed to be routed
through the files

lily/include/smobs.hh
lily/include/smobs.tcc
lily/include/small-smobs.hh
lily/smobs.cc
lily/undead.cc

So disabling the registration hooks should not affect too many lines.
I think just outcommenting the lines

  if (&Super::mark_smob != &Smob_base<Super>::mark_smob)
    scm_set_smob_mark (smob_tag_, Super::mark_trampoline);

in lily/include/smobs.tcc should disable all hooks.  At least there does
not appear to be any other call to scm_set_smob_mark in the code base.

> And are we supposed to include the BGC headers ourselves and issue
> GC_blah commands directly?

Oh no, just use the same mechanism as in Guile-1 (which is what we do
currently).  Using the hooks is seen as a second-class citizen, but
after years of bug reports and nagging, our tests stopped crashing
because of GC problems.

-- 
David Kastrup