Re: SunOS's xterm pb : again !

[mouse () Collatz McRCIM McGill EDU (der Mouse)]

> > > > [...ownership and permissions of /dev/ttyp*, as they affect
> > > > security of non-setuid xterm...]
> [...]

> The System V way of allocating ptys is really superior.

Which way is that?  The one that gave us pt_chmod, or was it pt_chown?
(Sorry, cheap shot.)

Seriously, speaking as an application author, I have been unable to
perceive a "System V way of allocating ptys".  I have one piece of code
that allocates ptys the Berkeley way, and it works with zero changes on
every Berkeley-based system I've encountered.  I have one more piece of
code for _each_ SysV variant I've managed to port pty-allocating code
to.  Each one seems to have invented its own way of allocating ptys,
all incompatible and all documented as being the only supported way
("other methods may not be supported in future releases" is a typical
phrase).  Sometimes, even, different releases of the same OS use
different and incompatible ways.

Whatever its technical flaws may be, the Berkeley way has the advantage
of being significantly less nonportable.  At least one SV-based system
I've seen has arranged for BSD-style pty allocation to work, or at
least claims to have; I didn't put it to the test.

As for relevance to bugtraq, well, when every vendor is inventing its
own way, sometimes re-inventing it for each release, guess what that
does to the bug density :-)

                                        der Mouse

                            mouse () collatz mcrcim mcgill edu