This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Improving --enable-libstdcxx-time
- From: Jonathan Wakely <jwakely dot gcc at gmail dot com>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: "libstdc++" <libstdc++ at gcc dot gnu dot org>
- Date: Wed, 29 May 2013 09:20:45 +0100
- Subject: Re: Improving --enable-libstdcxx-time
- References: <CAH6eHdQzDRwEYFrFqY0Yv8yCX+fKxcFjqz4GKaO6z+qpzmwOAA at mail dot gmail dot com> <CAH6eHdT160a5nfGGHiw3B0pU+toaT0rejAUahjxexBXhQ+N5_A at mail dot gmail dot com> <CAH6eHdQgRD74ichX2E2cv0RY8prY-f4-xNG_rOaz07n1ACH3iQ at mail dot gmail dot com> <20130528230324 dot E8EE32C05D at topped-with-meat dot com>
On 29 May 2013 00:03, Roland McGrath wrote:
> Version checks are pretty much always wrong. It should use an empirical
> check on whether you can link a program calling clock_gettime without
> passing -lrt.
In principle I agree, but IIUC we can't do link tests for libstdc++
when building a cross-compiler, and the rule is that the same
configure flags should produce the same configuration for
cross-compilers as for native, so we can't do link tests by default
ever ... so we can't make use of your clock_gettime changes unless
users explicitly request them, which they don't do (because they
probably don't know about it.)