Re: master 5c532fe303: Recommend that the user turn off memory overcommit

[Po Lu <luangruo@yahoo.com>]

Brian Cully <bjc@spork.org> writes:

> 	These features prevent Emacs from building, and I’d argue,
> constitute evidence of bugs in Emacs, as it’s operation is at odds with
> standard and necessary computer security. The wording in etc/PROBLEMS
> echoes this sentiment.

Whatever they might constitute, those changes might be required for
Emacs to build correctly.

> 	Your proposal crosses a line into how people should administer
> their systems even when Emacs functions as expected within the context
> of that system. ie, if I’m on Linux, overcommit is standard and I
> presumably know how to deal with it or am ignorant of it. In either
> event it is not the place of my text editor to tell me that memory
> overcommit is bad. There are countless opinion pieces on the web that
> will do that job just fine, should I seek them out.

None of this says that memory overcommit is "bad", it says that it's
detrimental to the correct functioning of Emacs.  (And "correct" here
means memory_full being called before Emacs is killed by the OOM
killer.)