Re: pdumper on Solaris 10

[Óscar Fuentes]

Eli Zaretskii <eliz@gnu.org> writes:

> Maybe so, but why is such a long wait a problem?  GC _works_, and
> works well.

Working on certain projects with lsp-mode is a miserable experience due
to all the random pauses.

My perception of the past week or two using igc is that those pauses are
much less jarring, if perceptible at all. I need more time to make a
definitive judgment, though.

As code edition evolves and Emacs is put on more demanding tasks the
limitations of GC become more obvious (and CPUs are not getting faster
anymore).

Apart from that, I'm convinced that there is quite a bit of evolutionary
pressure exerted by GC on the Elisp package ecosystem: something that
works too slowly or is too bumpy does not atract users and die. Others
may end devoting a lot of effort to optimize GC usage and when they
finally work "well enough" (for some generous interpretation) most
potential users already made their mind (flx.el is a paradigmatic case)
or the package author simply stops working on it, sometimes without
making the first release.

GC also diminishes the benefits of native-comp and other performance
enhancements: no matter how fast you make your Elisp execution engine,
the time taken by GC stablishes a hard limit.

But the "stop the world" mode of GC operation makes user experience
quite worse even if the total time to perform a task is smaller.