Boost logo

Boost :

From: John Maddock (jm_at_[hidden])
Date: 2003-08-18 05:31:44


> > My understanding is that Boost.Config should take care about these
issues.
> > Boost.Test rely on BOOST_HAS_SIGACTION flag. It should not be defined in
> > case if there is no support for POSIX interfaces. Could you report the
value
> > of that flag in case of compilation failures you are expiriencing.
>
> BOOST_HAS_SIGACTION gets defined to an empty string
> for como, "gcc -ansi -U_GNU_SOURCE" and for "icc -D__STRICT_ANSI__".
>
> Boost.Config uses _POSIX_VERSION to determine wether sigaction()
> is available. The presence of _POSIX_VERSION doesn't indicate
> wether the POSIX API has actually been enabled.
>
> If we want to use Boost.Config to take care of this then
> Boost.Config also has to check wether POSIX has been enabled.
> This would be a very tedious task. glibc uses a plethora of
> flags to enable POSIX, other implementations probably will
> also add some flags.

My gut feeling is that checking for _POSIX_C_SOURCE would probably be the
standard conforming way to handle this, but is likely to break on some
systems.

Could we not just add <define>_POSIX_C_SOURCE=200112 to the test library's
requirements?

On the other hand, I notice that the requirement to specify this, is only
for conforming C applications, nothing is mentioned about other languages
(if we're being picky about this).

John.


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk