Re: any objections to adding -D_PTHREADS in mk/pthread.builtin.mk and pthread.buildlink.mk

[Richard PALO <richard.palo%free.fr@localhost>]

Le 10/10/13 08:24, Martin Husemann a écrit :
> On Thu, Oct 10, 2013 at 07:40:44AM +0200, Richard PALO wrote:
> > Certain pthreads based applications specifically check for _PTHREADS
> > being defined, such as boost on solaris in which absence of this
> > CPPFLAGS issues this type of message:
>
> The compiler is supposed to define that when used with -pthread:
>
>     gcc -pthread -dM -E - < /dev/null | fgrep _PTHREADS
>
> and compare w/o -pthread. If your compiler doesn't do that, add it to a
> OS/compiler specific section, as Obata suggested.
>
> Martin
This works, so I'll go back to what I originally tried but figured that 
after so many years it would have already been dealt with there:
> richard@devzone:~/src/pkgsrc/mk$ git diff .
> diff --git a/mk/pthread.builtin.mk b/mk/pthread.builtin.mk
> index 733edc9..03810d5 100644
> --- a/mk/pthread.builtin.mk
> +++ b/mk/pthread.builtin.mk
> @@ -61,7 +61,7 @@ BUILDLINK_LDFLAGS.pthread=    # empty
>  # XXX
>  # XXX This should really be a check for GCC!
>  # XXX
> -BUILDLINK_OPSYS_SUPPORT_PTHREAD=       DragonFly FreeBSD Linux MirBSD NetBSD 
> OpenBSD
> +BUILDLINK_OPSYS_SUPPORT_PTHREAD=       DragonFly FreeBSD Linux MirBSD NetBSD 
> OpenBSD SunOS
>  .    if !empty(BUILDLINK_OPSYS_SUPPORT_PTHREAD:M${OPSYS})
>  BUILDLINK_CFLAGS.pthread+=     -pthread
>  BUILDLINK_LDFLAGS.pthread+=    -pthread


OKAY?