Re: [Emacs-diffs] trunk r117941: Default to stack objects on non-GNU/Linux, non-DOS_NT platforms.

[Eli Zaretskii <eliz@gnu.org>]

> Date: Thu, 25 Sep 2014 09:06:42 -0700
> From: Paul Eggert <eggert@cs.ucla.edu>
> CC: emacs-devel@gnu.org
> 
> Eli Zaretskii wrote:
> > I don't understand these sneaky practices, and I wish they'd
> > stopped.  How many times do you have to be told to revert your commits
> > before you understand that this is not how we do things around here?
> 
> The issue was briefly discussed, there were problems with MS-Windows 
> that I do not follow and cannot easily test, Dmitry played it safe and 
> left MS-Windows alone, and I followed his lead.  It was reasonable for 
> us to be cautious about enabling USE_STACK_LISP_OBJECTS on an untested 
> and potentially-problematic platform.  And it was reasonable for you to 
> enable USE_STACK_LISP_OBJECTS on MS-Windows once you installed further 
> changes to fix the problems and had tested the result.

The point is that whether to turn this on by default and for which
platforms was never discussed.  Dmitry asked for help in testing that
on Windows, and I provided my conclusions about that.  Another user
reported problems with clang.  That's it; no further discussions were
conducted.  What you say above does not (and cannot) refute this basic
fact.

This is not how this should be done.  We should discuss first, arrive
at some conclusion, and only then act.  This is a team project, and no
single one of us can hope to grasp all the effects and consequences of
any serious changes such as this one.

> the above-quoted remarks are unwarranted.

On the contrary, they were well deserved.

> By the way, I share Stefan's concern that the performance improvements 
> of USE_STACK_LISP_OBJECTS haven't been adequately demonstrated.

So do I, but that's another issue, unrelated to what I protested
about.