[ I've been letting these pile up for a day.  now i'm goign to answer
them in reverse order. ]

> >           I really want to be able to point my friends and associates
> > at a stable release as great as 1.0, and being able to let them just
> > go at it.  So, I guess what I'm saying is, can we make the next major
> > release a little less ambitious, and, say, not more than six months
> > from the previous release (1.0)?  I'm not asking you to lower your
> > sites -- simply to take the leaps in small increments.  I think it
> > would really help make it usable for the people who can't quit grasp
> > upgrading to current, especially in the later stages.  This has really
> > been my only beef with the NetBSD development, thus far.
> 
> Please excuse my bad english, but isn't it possible to give a 
> ''current' - monday tree' to someone who could build a distribution?
> Someone who is building a set of floppies for someone who wants to
> establish a new system? This may be a '1.001' and a '1.002' system.
> A system wich is not 'correct' but running.

In a word: NO.

If an auto manufacturer were designing a new car, and there was
great public demand for that car, would you rather the automobile
manufacturer start building and shipping them in quantity before
the safety analyses were complete, or would you rather they wait?

There's a _hell_ of a lot to do to build a 'distribution'.  First of
all, you have to coordinate between the various ports that are going
to be a _part_ of the distribution.  Then you have to branch the tree,
and start having the people responsible for the ports have people
alpha test them.  You kill bugs, eventually it becomes beta-testing.
You kill more bugs, and tag it as "final."  you have everybody build
binaries and kick them out the door.

You _must_ have a release be as well-tested as possible.  Why?
because it will persist nearly _forever_.  Almost a year after
it was released, people are still installing 0.9.  It'll be the
only thing available on some FTP sites for several months after
1.0 is released.

Who gets the bug reports for these releases?  In general, it's _not_
the people who made the release, it's netbsd-bugs@lamp.  It's a _lot_
of work to create a release, it _shouldn't_ be done without testing,
and it shouldn't be done too _often_, because that's a revision
management nightmare.

What you release today isn't worth a whit if it isn't a viable
platform _tomorrow_, because people are going to run it _then_.

> BTW. why is the sender of the mail-list e-mail a personal account and
> not the LIST? It is easier for me to save the mails to 'current' if 
> the sender also is 'current' or how the list is named today.
> It may be that 'you' only save the messages of some people from
> the list and then under their names. But I want to save the messages
> under the list name.

Who is the sender of the message?  is it the mailing list, or the
person who sent the message?  (obviously, it's the latter; it
gets sent _to_ the mailing list.)

There's also the fact that if the message is "From:" the mailing
list, it's bloody difficult to reply to the real sender!  I certainly
hope you believe that not every reply should be made to the mailing
list...  8-)

Besides, if you're using a 'reasonable' mail system, it's just as
easy to sort on 'To:' as it is on 'From:'.  In general, i sort
mail that i consider 'bulk' by To primarily, then Cc, then From.


cgd

------------------------------------------------------------------------------