This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Reserve __is_trivially_relocatable for a builtin?
- From: Jonathan Wakely <jwakely at redhat dot com>
- To: Ville Voutilainen <ville dot voutilainen at gmail dot com>
- Cc: libstdc++ <libstdc++ at gcc dot gnu dot org>
- Date: Mon, 4 Feb 2019 12:17:54 +0000
- Subject: Re: Reserve __is_trivially_relocatable for a builtin?
- References: <alpine.DEB.2.21.1902020004580.9809@stedding.saclay.inria.fr> <CAFk2RUZtGV=SMqCWoXQhagTsvTymOgJGr6mx7rR482oCW47c6A@mail.gmail.com>
On 02/02/19 01:25 +0200, Ville Voutilainen wrote:
On Sat, 2 Feb 2019 at 01:12, Marc Glisse <marc.glisse@inria.fr> wrote:
Hello,
in https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87106#c15 , Arthur O'Dwyer
suggests that the name __is_trivially_relocatable would be a good name for
a future builtin, and we should use something else for this class. Does
anyone have an opinion on whether we should rename and what other name we
should use? I am tempted by __is_trivially_relocatable__,
Inconsistent with everything else in libstdc++.
_Is_trivially_relocatable
This would be fine.
or ___is_trivially_relocatable ...
Definitely not this :-)
We have a mixture of __do_{your chosen name goes here} and {your
chosen name goes here}_impl
throughout the library. There's no real chosen policy for it. I have
strong objections to playing
games with the amount of underscores, library maintainers don't see
them. :) My suggestion
is to rename the current one to __is_trivially_bitwise_relocatable.
That would be OK too.