In message <963cd04f05072516054fb0ff38@mail.gmail.com>, lama writes:
>2005/7/25, Richard Rauch <rkr@olib.org>:
>> Hm...I always thought that pkgsrc *did* do a forwards/backwards
>> update of all necessary packages.  This is why it is so painful to
>> do "make update": > I've seen the kind of message that you describe, but it'
>s rare.  My
>> memory is that it happens only when I cannot do a successful
>> "make update" on the package itself to get the desired update.
>> 
>> Is it possible that you have altered /etc/mk.conf in some way that
>> affects this?
>> 
>
>I still  don't understand what's wrong with pkg_chk why aren't all of
>You using it ? Am I missing something ?
>
>
>NAME
>     pkg_chk - check, and optionally update, installed packages
>
>SYNOPSIS
>     pkg_chk [-aBbcfhiklNnrsuv] [-C conf] [-D tags] [-L file] [-P path]
>             [-U tags]
>
>DESCRIPTION
>     pkg_chk verifies that the versions of installed packages matches those in
>     pkgsrc, optionally adding missing or updating mismatched packages.
>
>     It can verify all installed packages, or check against a set of packages
>     based on the hostname, architecture, and other characteristics.
>
>     When updating packages that depend on each other pkg_chk will skip depen-
>     dent packages to reduce unnecessary rebuilding.
>

There's also this statement in BUGS:

     If both package a and package b are to be updated, and a depends on b,
     pkg_chk will correctly skip the update of a.  However, if a depends on b
     and c, and all three are marked for update, pkg_chk will update b and c
     in two separate passes, resulting in unnecessary rebuilding of a (and
     potentially other packages).

It's not clear to me how that plays against the last paragraph you 
quoted.

		--Steven M. Bellovin, http://www.cs.columbia.edu/~smb