Default Boost Versions for "R", may have to be 1.50

[Dmitrijs Ledkovs]

On 27 October 2012 00:12, Dmitrijs Ledkovs <dmitrij.ledkov at ubuntu.com> wrote:
> On 17 October 2012 17:51, Scott Kitterman <ubuntu at kitterman.com> wrote:
>> We had a bit of a discussion about this on #ubuntu-motu yesterday and, among
>> those discussing it (who are the ones who usually do most of the work on a
>> Boost transition), we concluded staying in line with Debian and keeping 1.49
>> as the default Boost is what we should do for "R".  As with Quantal, Boost1.50
>> will be available in Universe for developers or packages that need the newer
>> features, but it won't be in Main and should not generally be used for
>> packages unless it's really needed.
>>
>> The good news is that means there there is nothing that needs doing for Boost
>> before "R" opens for development.
>>
>
> Now the bad news.
>
> The new eglibc 2.16 defines TIME_UTC macro which is part of C11 and
> clashes with boost::TIME_UTC which means boost1.49 does not compile
> any more.
> (anything using boost/thread/xtime.hpp template or explicitly using
> boost::TIME_UTC)
> Luckily this is fixed in boost1.50 by means of renaming the macro to
> boost::TIME_UTC_
> There are rdeps which actually use boost::TIME_UTC macro all of them
> need to transition to boost1.50 and boost::TIME_UTC_
>

If it was not obvious from the doko's archive opening email, for the
record here is the update on boost.

We applied a patch to undefine TIME_UTC in boost1.49 package, if
TIME_UTC is defined. This preserves existing boost1.49 api and keeps
it buildable with the eglibc2.16 in raring. This workaround means that
we are keeping boost1.49 as the default boost as per original plan.

Regards,

Dmitrijs.