Re: Update of wip/libressl to 2.1.6

["Kamil Rytarowski" <n54%gmx.com@localhost>]

> Sent: Sunday, March 29, 2015 at 7:42 PM
> From: "Kamil Rytarowski" <n54%gmx.com@localhost>
> To: "Joerg Sonnenberger" <joerg%britannica.bec.de@localhost>
> Cc: pkgsrc-users%netbsd.org@localhost
> Subject: Re: Update of wip/libressl to 2.1.6
>
> 
> 
> > Sent: Sunday, March 29, 2015 at 3:52 PM
> > From: "Joerg Sonnenberger" <joerg%britannica.bec.de@localhost>
> > To: pkgsrc-users%netbsd.org@localhost
> > Subject: Re: Update of wip/libressl to 2.1.6
> >
> > On Sun, Mar 29, 2015 at 05:06:07PM +0200, Kamil Rytarowski wrote:
> > > I had a closer look at this issue and it's a problem how GNU Autotools discover libc functions.
> > 
> > Please be correct. Which way to probe for a function did you use?
> > autoconf offers a number of options here, from plain "does something
> > with the same external name as this symbol exist" to "here is an
> > expression, can it be compiled and linked". It is a common issue to use
> > only the former form, when the latter produces much better results, but
> > needs more verbosity.
> > 
> 
> In configure.ac there is:
> AC_CHECK_FUNCS([reallocarray])
> 
> I was testing this discovery in a dummy project too and it claimed that
> reallocarray() (on my -current box) is available.
> 
> Sorry for seeing your reply after my next post, I think we need to learn
> how to handle the _OPENBSD_COMPAT layer and document it.
> 

The issue with reallocarray() should be fixed now in wip/libressl.

Nobody opposed so I added:
# Added for 2.1.6 - remove for next stable version as it's fixed upstream
CPPFLAGS.NetBSD+=       -D_OPENBSD_SOURCE