Re: [gpsd-dev] regresssion from 2011: git head fails to build on NetBSD

[Eric S. Raymond]

Greg Troxel <address@hidden>:
> 1) endian.h is unconditionally included without testing for where it is.
> On NetBSD it's sys/endian.h.  AFAICT POSIX does not specify this, so a
> configure-time test is necessary.

Fixed in git head.  I've implemented a test in Python that is not
dependent on compiler magic.

>  Is the current code tested on any BSD?

I don't know.  Chris Kuethe hasn't been around much lately.

> 2) gpsd.h uses va_list without having included stdarg.h.  It seems
> that's required by POSIX but I'm not 100% sure.  Adding stdarg.h allows
> the compile to succeed.

Fixed in git head.
 
> 3) curses/curses_dll.h was not found, even though it's in $prefix, and
> I was unable to find a --with-ncurses=prefix argument to scons from
> reading the help output (to cause the corresponding -I and -L/-R
> arguments).

curses/curses_dll.h?  What?  Where does GPSD use that? No grep hits...
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>