Re: Eager garbage collection

[Eli Zaretskii]

> From: Spencer Baugh <sbaugh@catern.com>
> Date: Tue, 17 Nov 2020 19:20:47 -0500
> 
> Stefan Monnier <monnier@iro.umontreal.ca> writes:
> >The time to do a GC doesn't depend on the amount of garbage so much as
> >the size of the heap, but I think I like your idea.  Better yet: keep
> >track of the time that each GC takes and use that as a guide for the
> >idle-time delay.
> 
> OK, I added an exponential-moving-average to track the time that GC
> uses, and use it as the idle-time delay, with the idea that if we
> reach idle time X, the naive best estimate of the total idle time is
> 2*X. See following patches.

I'm trying to understand whether we should enable this by default or
not.  The decision should depend on the overall effect of this
feature, and on whether it has any downsides.  IME, changes in GC
details frequently have unintended consequences.  Can you present some
evidence of the effect of using the feature, either measurements or
even anecdotal evidence?

More generally, perhaps we need some tools to measure the effect of
this (e.g., how many GC's happened during some rime interval, how much
time each GC took, how large is the VM size of Emacs, etc.  Then we
could ask people to present numbers with and without the feature, and
make the decision based on that.

Thanks.