update stream policy
ducasse at iam.unibe.ch
Sat Jan 17 21:29:06 UTC 2004
On 17 janv. 04, at 21:58, Avi Bryant wrote:
> On Jan 17, 2004, at 12:38 PM, ducasse wrote:
>> But I was not talking about package code with a maintainer. I was
>> talking about Kernel OrderedCollection, ClassBuilder,
>> SystemChangeNotifier...the core.... the deep core....
> Well, the question for me is, who is the maintainer? Or are you
> saying there isn't one?
Nobody and everybody is maintaining the image. Who is responsible?
Nobody but everybody care. That's why any people doing development on
squeak should review fix and bug for the image because this is
***our*** asset. tests are documentation
> Does the same person maintain the OrderedCollection code that
> maintains the OrderedCollection tests?
Not necessarily but if there is test for OrderedCollection, I review it
and approv it if this is ok and the OrderedCollection maintainer does
not have to worry. In fact, I would loved like andreas did that people
submit tests for the core of Squeak, this would
help us ****a lot****. tests are documentation
> ClassBuilder and the ClassBuilder tests? If so, then ok, put the
> tests in the image (until we have it together enough to maintain even
> the core elements as discrete packages),
Exactly my point. tests are documentation
> but give the designated maintainer full commit privileges - no
> approval needed for them to get changes into the stream. At least
> that way we don't get bogged down so much in the BFAV.
But who has problem with too much tests in BFAV.....be serious there is
nobody except andreas, boris and me submitting tests (sorry for the
only ones I know you are cool submitting tests). So the harvesting of
tests for the image is NOT a problem at all. tests are documentation
Right now marcus was the maintainer of these tests but I did not want
to ask him to update all the time the package
because marcus has better stuff to do like improving the compiler.
tests are documentation
> My sense is that right now there is a clear group of people that are
> maintaining these tests, separately from whoever maintains the core of
> the image (ie, everybody and nobody), and could most easily keep
> maintaining them if they were a separate package independent of the
> image. But I could be mistaken about that in this instance, and maybe
> that's what you're telling me.
Yes this is was I'm telling to you. as a KCP person, I'm maintaining,
supporting, encouraging tests for the basic image, marcus
was collecting them but nobody saw them so we were not having enough
awareness that the tests are usefull (again an education point, if
everybody would produce tests and squeak would be fully covered by
tests, I would not care that much but everybody would be convinced that
tests are the only documentation we can afford to have :)) So for us
this would be much simpler
we should reduce the entry level for the three test-infected guys out
there that are thinking about the image. tests are documentation
So I was not saying that tests for any package should go in the image.
Only the code that is in the image. Now as soon as we chainsaw it then
the tests should leave to live happily with their code on a package
somewhere. For me the fewer in the image the better, but each time KCP
will modify something we will not be fixing the packages out there. So
their maintainers should start now writing tests or running the system
until they get a deprecated warning and fix it. Which is ok for me. the
smaller the image the easier the KCP changes are, less mess, less
trouble more fun. :) tests are documentation
But now I simply want to have to spent less energy to maintain the
tests that I take care of and that are related to the image.
Also I do not want to see core stuff been blocked because they contain
tests. Marcus in the past spent time separated tests from code and this
is a lost of time. tests are documentation...
Stef tests are documentation
I'm trying a new technique for subliminal messages in email but I'm not
sure it really works :)
So avi we agree at the end. I was not asking for the moon. Just a way
to help us sparing energy and increasing visibility of
More information about the Squeak-dev