This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: configuring out old ABI


On 2018-06-06 10:30, Jonathan Wakely wrote:
> On 06/06/18 10:17 +0200, Rasmus Villemoes wrote:
>> I naively attempted
>>
>>  --disable-libstdcxx-dual-abi --with-default-libstdcxx-abi=cxx11
>>
>> and was a little surprised to find out that the second option is
>> ignored, and I got the cxx98 ABI (only). Of course, it's documented if
>> one reads the fine print in libstdc++-v3/acinclude.m4
>>
>> dnl --disable-libstdcxx-dual-abi will use old ABI for all types.
>>
>> or the logic a little further down that only honours
>> default-libstdcxx-abi when x$enable_libstdcxx_dual_abi = xyes.
>>
>> Is there any chance of having a configuration combination completely
>> disabling the old ABI?
> 
> Not currently, this is
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65762
> 
> If/when https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83077 gets
> implemented you could get it via
> --enable-symvers=gnu-versioned-namespace (but that also changes the
> ABI to be completely incompatible, instead of just dropping the
> non-cxx11 parts).

Thanks for the quick reply. In my case, I only use the static libgcc and
libstdc++ (I'm targeting vxworks), and don't even know if I could build
a .so, so it seems that the logic in acinclude.m4 would complain loudly.
How would/does symvers affect libstdc++.a?

Rasmus


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]