[boost] Re: Boost 1.32 vs. tru64cxx65

From: Markus Schöpflin (markus.schoepflin_at_[hidden])
Date: 2004-11-23 11:00:34

Richard Hadsell wrote:
> Markus Schöpflin wrote:
> 
>> The problem is Digital UNIX V4.0F. It doesn't have inttypes.h. Try to 
>> remove the check for __osf__ in boost/cstdint.hpp:85 and see what 
>> happens.
> 
> I did that, and it worked fine.  I thought someone might like to know 
> that it could use some kind of configuration macro.

If you happen to know a way to distinguish the two versions on preprocessor 
level, a fix would be easy.

>> But be warned, no regression tests were run for Digital UNIX V4.0F, 
>> only for V5.1, so you are on your own.
> 
> I understand the risks of using ancient systems.  I won't complain about 
> new or changed Boost libraries that no longer build on them.  I'm hoping 
> we can retire them in the near future.
> 
> I'll check out John M's suggestions.  While I do that, I wonder whether 
> there is any bjam build option for tru64cxx65 that selects any 
> configuration options distinguishing Tru64 4.x from 5.x.  If not, 
> probably the easiest solution for me is to just take your suggestion.  
> (I have no Tru64 5.x systems.)

I don't think that currently there is any distinction at all between 4.x 
and 5.x, neither in boost nor in boost.build or in the toolset.

What really amazes me is that you can use V6.5 on Tru64 4.x, I always 
thought that V5.x was the latest compiler toolset for Tru64 4.x.

Markus