cvs commit: src/sys/sys mbuf.h src/sys/kern uipc_mbuf.c uipc_syscalls.c src/usr.bin/netstat mbuf.c src/lib/libc/sys sendfile.2

[Ken Smith]

[ Let me take a bash at this, since at this point if I don't fully
understand this I'm in trouble.  If I miss the target feel free to
comment. ]

On Fri, Jun 18, 2004 at 01:20:31AM +0700, Max Khon wrote:
> Hi, Alfred!
> 
> On Thu, Jun 17, 2004 at 10:38:54AM -0700, Alfred Perlstein wrote:
> 
> > MFC'ing a change to netstat -m's output is not acceptable.
> > 
> > This will break scripts.
> > 
> > Please back out the netstat change.
> 
> No problems, but can you tell me which scripts or programs
> has become broken after this change?

The scripts that are now potentially broken don't necessarily need to
be in the base system.  They could very well be scripts Alfred wrote
and uses himself for his own purposes.  Once a branch goes into -STABLE
status we try to not make 'user-visible' changes to it unless they are
bugfixes or do not alter the way something had worked.  Your change
alters the way an existing flag works.  That would be acceptable if,
for example, we had claimed the output with that flag was posix compliant
and due to a bug we were not truly posix compliant.  That's a bugfix.
But if it's not a specific bugfix we try to not change the semantics
of a command's output once a branch is -STABLE which RELENG_4 is.  It's
OK to add new flags because a user would not be relying on those in their
own scripts.  It's also OK to only add something to -CURRENT and not
MFC it to the -STABLE branch if you feel it would break the above
criteria.  And being able to make user-visible changes to things is
exactly why there is a distinction between the primary development
branch and the -STABLE branch, and why the 5.X branch shifting to
-STABLE at 5.3 is considered a big deal.

> > Why are sfbuf/sendfile stats being put under mbuf stats?
> > 
> > This is bogus, please use a different flag instead of cluttering
> > the mbuf stats.
> 
> This is how things are implemented in CURRENT.
> I do not think that implementing it differently in RELENG_4 makes sense.

Per above - if it were an invisible change (no change to what a user
could see) then it's OK to implement things the same in CURRENT and
RELENG_4.  But countless things are now implemented much differently in
CURRENT versus -STABLE because of what I describe above.

-- 
						Ken Smith
- From there to here, from here to      |       kensmith at cse.buffalo.edu
  there, funny things are everywhere.   |
                      - Theodore Geisel |