QA in Squeak (was: Re: [Bug][Fix] Setting
copy/paste-keypreference under Windowsdoesnot work)
Marcus Denker
denker at iam.unibe.ch
Wed Dec 8 17:06:43 UTC 2004
Hi,
Just some general remarks...
1) In the current "process", there is no automatism for a changeset
that is
tagged "et works for me" to end up in the update stream. The decion of
that is in the hand of the harvester actually approving the changeset.
e.g. there is a lot of stuff on BFAV that hat [et] but it's not in the
stream.
And those harvesters do make errors. And doing those errors, especially
those
of the quality "I should have known that" do not feel good, I can tell
you. (It's quite
embarrassing).
(disclaimer: i neither reviewed nor approved the .cs in question ;-)).
2) I am completely with you that certain things need real expert
review, and I think
that overall, the whole thing didn't work too bad.
So, expert review is need. But what happens if this expert review just
*never* happens?
Everthing related to harvesting/reviewing has been *completely*
abandoned by most of
the real good people. Even explicilty asking: "We need a reviewer for
EToy related things",
"We need a reviewer for networks stuff" "we need a reviewer for mophic
stuff" has
been completly IGNORED. This may work for a year, or two. But not
three.
I call this "doing nothing is not an option", or the "horror vacuui" of
Open Source
development: At some point the pressure just gets to big and something
happens.
Back in the SqC times, the small parts of the many community
submissions that made
it were exactly that bit needed to calm the pressure. Not perfect, but
good enough. But
then came a time when *nothing* happend.
And if then, forced by the vacuum, something happens, it might be that
it is not that good,
not exactly that direction that it was before. For the direction, you
need to manage the
community more activly. For the quality, you need to make sure that not
everybody good
abandons the ship.
Marcus
More information about the Squeak-dev
mailing list
|