Jonathan Perkin <jperkin%mnx.io@localhost> writes:
* On 2023-05-23 at 15:20 BST, Tobias Nygren wrote:
On Mon, 22 May 2023 11:16:29 -0500
Jason Bacon <jtocino%gmx.com@localhost> wrote:
I certainly could, but I suspect upstream will argue that
clang++ -x c file.c
is perfectly valid, and we'll have to keep the patch around
indefinitely. Even if they agree to change it, this may pop up in other
packages, and someone will have to spend time solving this problem again.
It is not that easy to patch since you have to hook into the boost-jam stuff.
I could reproduce on latest Ventura (aarch64). The -std=c++14 comes
>from meta-pkgs/boost/Makefile.common's USE_LANGUAGES.
The least intrusive quick-fix seems to be to undo this platform conditionally.
Given this was only added to support NetBSD 8.0/i386, I'd rather see
us add a hack for that platform to add to USE_LANGUAGES, then at least
we're being explicit about what platform we are fixing (and can remove
at a later date if necessary).
But is this a hack, or was it right? Does boost document that it needs
C++14? Does it fail if built with a compiler that supports C++11 and
not C++14?
I can see the point that on later systems the compiler supports C++14 by
default and thus a --std=c++14 is not needed, but that feels like a
workaround for not settting --std!