Hi all!

Chris Muller <afunkyobject at yahoo.com> wrote:
> May I agitate this thread with intentions of good?

:)
 
> In my personal experience, the actual submission of changes/fixes has been
> difficult and laborious.  After the actual submission, some of them further
> required continued hard lobbying/nagging/reminding, etc.  Even after hours of
> work, many never made it.
> 
> With such a high "cost" of getting a change, it really emphasizes what Lex
> seems to be complaining about..  That the process does not support small,
> simple, safe fixes.  What if someone merely wants to correct a class or method
> comment?  There's no way that's worth going through the pain of the current
> process, not only for the submitter but for the harvester, and that's a shame
> because lots of "baby-step" improvements can contribute a lot to the maturity
> of a system.

Just so you know - we have recently talked quite a bit about changing
the process.

The current status is that 3.7 will be out very soon (Doug is working on
it), the m17 stuff is soon hitting the 3.8 stream and that will possibly
create lots of issues to take care of.

When 3.7 is out I think we should try a few things:

- First of all I want the TFNR picked up, by starting "handing out" a
few selected areas of the image to some maintainers that have stepped
forward and are interested. Note - this does NOT include splitting out
stuff, just staking it out and assigning it to maintainers. They can
then use BFAF/update stream in the beginning.

- Secondly we have been thinking in terms of maintaing parts of the
image using Monticello directly instead of using the update stream. Then
we should probably have "commit rights" for the experienced people and
letting them commit freely under at least alpha and beta.

Eventually the goal is for the image to consist of separately maintained
packages and the update stream would be replaced with a dependency
driven upgrading engine that can upgrade the image in a controlled
manner.

Well, I am pretty busy with SM. To get the above balls rolling we need
people to pick them up.
To be as concrete as possible, we need some to:

1. Figure out how the PackageInfo registry in the image should be/work.
Avi? It was a while since I looked at PI, but we need to be able to
"push" PIs into the image using the update stream. To be even more
concrete - I have a willing maintainer for Collections :) and it would
seem like a good place to start as any. So can we get a PI for
Collections into the stream? :)

2. Setting up Monticello so that we can start testing how it works using
it for image maintainance.


regards, Göran