Re: MPS: User GC customizations

[Eli Zaretskii <eliz@gnu.org>]

> From: Gerd Möllmann <gerd.moellmann@gmail.com>
> Cc: Pip Cet <pipcet@protonmail.com>,  yantar92@posteo.net,
>   emacs-devel@gnu.org,  eller.helmut@gmail.com
> Date: Thu, 04 Jul 2024 19:01:35 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> From: Pip Cet <pipcet@protonmail.com>
> >> I'm really confused. Can you explain what you think the difference
> >> is?
> 
> FWIW, I'm also confused.
> 
> > I just did, twice.  If that still doesn't explain it, then I guess my
> > explanatory talents betray me, and I don't know what else to say.
> >
> >> The variable's value is a number. If the variable is zero, we don't
> >> tell MPS we're idle when we're idle. If it's nonzero, we tell it
> >> we're idle (when we are), and how much time it's okay to spend in GC.
> >
> > The number can only tell one thing, not two.  AFAIU, it tells the
> > latter: "how much time it's okay to spend in GC" once GC started due
> > to idleness (and not some other reason).  Gerd said something
> > different: that this number "gives MPS notice" that Emacs is idle.
> > (How can a number "give notice" I don't understand even in principle.)
> 
> By saying that 0 has a special meaning.
> 
> The "once GC started due to idleness" confuses me. We don't "start" GC
> due to idleness.

I didn't say "we" start GC due to idleness, I said that GC starts due
to idleness.  Not necessarily by ourself.

> We allow MPS to do something concurrently, say one or more
> increments in the sense of its incremental GC algorithm. And "GC
> runs" all the time.

You are saying that the whether Emacs is or isn't idle is not relevant
at all to the effect of this variable?  Then why does MPS
documentation talk about "allowing the client program to make use of
idle time to do some garbage collection"?  And why does the doc string
of igc-step-interval itself mention "GC when Emacs is idle"?

IOW, let me turn the table and ask what does "idleness" have to do
with this variable?