Working Together (was: re: newbie question (...)) [LONG]

David Farber dfarber at numenor.com
Wed Jul 14 01:59:26 UTC 1999


just to throw in my 2 cents: i really like Dwight's extension of Dan's idea.
i think this would allow Squeak Central to "get out of the way" in certain
cases. i am thinking specifically about exceptions here. this should be a no
brainer. everybody wants it. people want it bad enough that they are rolling
their own. we need to pick one and run with it. even if none of them are
perfect, i think the interest and willingness to contribute is there to make
it perfect. so if Squeak Central were to commit to exceptions in v3.0, throw
someone's implementation into an image, mark it v3.0-CURRENT i know i for
one would be thrilled and would certainly want to contribute to bringing it
to 3.0-RELEASE.

david


At 07:51 PM 7/13/99 -0500, you wrote:
>Dan Ingalls wrote:
>[............]
>> Sometimes I get too busy to issue updates.  Most often, though, I am
>> actively planning how to shield the less adventurous citizens of
>> SqueakVille from the effects of dangerous and incomplete cycles of
>> change.  There have been a couple of times since the last release when I
>> could have issued updates with some confidence, but they happened also
>> to be times when I was busy.
>[............]
>> > What's worse is that the Disney team have their own internal update stream
>> > that we are not privy to. Recently, this resulted in a lot of duplicated
>> > effort by people on the list, fixing a bug that was already well and truly
>> > fixed. Now I realise that the Disney team are probaby working flat out on
>> 
>> While I dislike the pejorative allusion of the term "privy", I have to
>> agree that this is a problem.  Fortunately I think there is a simple
>> solution.  In the past we have been reluctant to release updates at
>> various junctures, in order to shield newbies from possible
>> instabilities.  What SHOULD happen is this:  when we have that feeling,
>> that is when we should declare a new version number.  Newbies would not
>> "see" any later updates, but any serious Squeaker could advance his
>> version and track the latest changes.  We at Squeak central can freely
>> broadcast everything hot off the press, knowing that only the
>> self-professed test pilots will be affected.  All the necessary
>> mechanism is already present in the update system.
>
>I like this suggestion a lot. The FreeBSD project uses a set of terms we
>could adapt rather nicely to cover just this sort of structure. I will
>use the terms as we could use them (the exact FreeBSD definitions are
>close but not quite the same). CURRENT: the latest, hot off the press
>code -- perhaps minimally tested or only tested in a limited fashion;
>this is where new subsystems are introduced, where new code is tried
>out; may go through several minor version numbers in the process; whole
>portions of the system can change in wholesale fashion; portions may
>actually be broken temporarily from time to time [what you get depends
>on when you grab it]. Of course, Squeak Central might not want _quite_
>that much of a bleeding edge CURRENT version -- but the understanding
>would be that things can get excessively interesting from time to time
>in the CURRENT path. STABLE: what CURRENT becomes as feature sets
>stabilize and biting bugs are smacked -- somewhere along the version
>scale between alpha and beta designation - updates are primarily to fix
>bugs with only minor enhancements or new features; feature set is
>essentially fixed. RELEASE: what STABLE becomes after a bit of time and
>use and bug fixing and polishing - should only see minor bug fixes
>released; feature set is fixed.
>
>Thus we could have (for example) a Squeak v3.0-CURRENT, a Squeak
>v2.6-STABLE, and of course Squeak v2.4c-RELEASE. 
>
>I think a number of us would *love* to be able to truly keep current
>with the update stream and have a little input into (or at least
>knowledge of) what is being done. The Squeak Team gets more eyes to look
>at their code, more people to hammer on it to find bugs even earlier,
>contributed bug fixes, people to do some of those additional little
>things that you would like done but cannot find the time.... It would
>also create a tighter sense of communication between Squeak Central and
>the rest of us - we will know which way the wind is blowing and be able
>to either adjust to it or express our reservations rather than have
>essentially complete subsystems dropped in our laps that we never knew
>were in the works. Some of Squeak Central's TODO items might get done by
>other interested parties, if they were aware of what some of your TODO
>items are.
>
>Just some thoughts.
>
>-- Dwight
>
>

--
        j. david farber
    oo architect+mentor
numenor labs incorporated
in sunny boulder colorado
    dfarber at numenor.com
        www.numenor.com





More information about the Squeak-dev mailing list