bug#40407: [PATCH] slow ENCODE_FILE and DECODE_FILE

[Eli Zaretskii <eliz@gnu.org>]

> From: Mattias Engdegård <mattiase@acm.org>
> Date: Sun, 5 Apr 2020 17:03:49 +0200
> Cc: 40407@debbugs.gnu.org
> 
> > I don't think I understand your line of reasoning here.  I don't think
> > GC is relevant, and as long as we are talking about file names, the
> > first null byte terminates it even though the Lisp string's length
> > could be larger.
> 
> It is stated as a reason in Fexpand_file_name for working on copies
> of strings; see comments therein.

That refers to code that keeps C pointers into string text.  This is
not our case here: we are talking about Lisp strings, not C pointers
into them.

> > I hope you now agree with me that we should not do this.  The default
> > should stay NOCOPY = false, and any caller that wants otherwise must
> > explicitly request that by calling code_convert_string.
> 
> I disagree -- if the callers handle the situation safely, there is no reason not to to do the change, saving some consing. We do this sort of code improvement all the time; nothing special about this one.

"If the callers handle the situation safely" is not a trivial
condition.  The programmer will more often than not be unaware of this
subtlety, and may not write such safe code.  Moreover, the callers
will have to handle it safely in the future, or be sure to insist on a
copy if not, and these two macros don't give the callers knobs to
request that.