Re: Eager garbage collection

[Eli Zaretskii]

> From: Andrea Corallo <akrl@sdf.org>
> Cc: sbaugh@catern.com, emacs-devel@gnu.org
> Date: Wed, 18 Nov 2020 15:30:24 +0000
> 
> > AFAICT, the basic idea is very different, so the results will probably
> > also be different.
> 
> My understanding is that they both:
> 
>   - Run the normal GC with a threshold X1
>   - Run after an idle time t GC with threshold X2, where X2 < X1
> 
> This patch compute X2 in terms of X1 using
> `gc-opportunistic-eager-factor' while in gcmh one specify
> `gcmh-low-cons-threshold' and `gcmh-high-cons-threshold' but
> conceptually looks the same to me.
> 
> Am I wrong?

The devil, as usually, is in the details.  What is radically different
between the two is the value of X1.  In the gcmh.el case, the value of
X1 is alarmingly large, something I keep warning people against for a
long time.  I don't want us to adopt such dangerous practices.  And
using that large value is central to the idea of gcmh.el, so if we
lower it significantly, I wouldn't expect it to provide significant
benefits.