Re: FYI: POSIX update - SIGWINCH and 'struct winsize' (etc) to be added

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Sun, 22 Oct 2017 19:47:27 +0000
    From:        David Holland <dholland-tech%netbsd.org@localhost>
    Message-ID:  <20171022194727.GA27523%netbsd.org@localhost>

  | Which? (i.e. which part of the logic I offered yesterday has a wrong
  | premise?)

That it is OK to break old sources doing things the way that it has been
(and for that matter, still is) documented that it should be done, just
because we now have a new way that is better.   It isn't OK.  A long
time after we stop telling people the old was is OK, it might (and only
might) be OK to break the old interface, but that is not now.

Kamil - I am not going to reply to your most recent message, as you're
getting to questions to which I have no real answers (and no real experience
from which to form rational opinions) - this is why I am asking here, to
try to discover what will work, and what will not..

But with (I think) some of what concerns you in mind, I have another
potential idea.

Could we perhaps use Christos' method, with the addition of modifying
<sys/ioctl.h> so that instead of just

#include <sys/ttycom.h>

it instead has ...

#if defined(_NETBSD_SOURCE)
#include <sys/ttycom.h>
#else
#define _NETBSD_SOURCE 1
#include <sys/ttycom.h>
#undef _NETBSD_SOURCE
#endif

?

If that's acceptable (and works, I haven't tested it yet - I have all the
other proposals) then I think it will be every bit as good as adding a new
header, without the costs.

kre