Hi!

I'm going to remove pkgsrc/mail/postfix/patches/patch-ac
and pkgsrc/mail/postfix-current/patches/patch-ac unless someone
can say why I really shouldn't do it...

These are what I got from Wietse Venema when submitting our pkgsrc
patches to him:

----------------------------
The code compiles warning free on NetBSD 2.0 and 2.1. I have not
adopted the patch that modifies vstream objects. It fixes the wrong
problem. No code must access the I/O time stamp before I/O timeouts
are actually enabled.

As documented:

/*      vstream_ftime() returns the time of initialization, the last buffer
/*      fill operation, or the last buffer flush operation for the specified
/*      stream. This information is maintained only when stream timeouts are
/*      enabled.

The field is updated first when the object is initialized to have
its I/O timeouts enabled. Code that accesses the I/O time field is
broken.

Changing this inulaterally on one platform may break future code.

	Wietse
----------------------------
This was fixed before Postfix 2.2, by not jumping into a non-existent
longjmp context.

The patch may break future code that expects that iotime
is zero after vstream_open.

	Wietse
----------------------------