An Improved Harvesting Process (was: Re: release prioritization)
Hannes Hirzel
hannes.hirzel.squeaklist at bluewin.ch
Sat Mar 1 12:23:40 UTC 2003
Doug Way <dway at riskmetrics.com> wrote:
>
> Daniel Vainsencher wrote:
> >
> > ...
> > [Status - everyone should know where each release stands. Could be
> > solved with a bug/issue DB]
> > You're right. We need such a tool, but - we need more urgently a fix
> > tracking tool, to make sure we don't lose work that people submit -
> > fixes are higher priority than bugs, because people become more upset
> > when you lose their changesets than when you ignore their bugs.
> >
> > We're planning to make such a tool, but we haven't made much progress in
> > that, either. We could really use some help from the community with
> > either of them.
>
> I put together a swiki page awhile ago (which I just updated a bit) with some
> initial thoughts on how such a tool might work:
>
> http://swiki.squeakfoundation.org/squeakfoundation/86
>
> Feel free to add comments at the end of the page. I'm planning to start
> working on this as soon as 3.4 is released, but if anyone else is interested
> in helping out with this, that would be great.
>
> - Doug Way
I have posted the following comment on
An Improved Harvesting Infrastructure
http://swiki.squeakfoundation.org/squeakfoundation/86
[Hannes Hirzel, 1-March-03]:
I do not think that the UI is the main problem. Even doing some manual
editing work on a swiki table is not really a time consuming thing.
The main problem is the process of assessing the quality of the fixes.
And this is directly linked to the availability of SUnit tests for them.
As a concrete next step I would suggest approaching the problem from
three directions
!! Harvesting process
Tentative description of actions for Enhancing the bug and fix reviewing
process
! Use case for harvester champion
The gsug archive contains the fixes http://swiki.gsug.org:8080/SQFIXES/
# Write a little script that loads all the [Fix] tagged changeSets from
the gsug archive into CodeBrowsers of a harvester champions flagship
image.
# run automated queries on the codebrowser to find out which class
categories are affected.
# generate sublists of groups of fixes - Smalltalk code that will load
the fixes for other harvesters.
! Use case for general harvesters
# Run one of the sublists to load a bunch of code browsers with fixes
# Have a quick assessment on the importance of the fixes.
# Do a table entry in a swiki table describing the fix and assigning it
a status.
# Ask the authors and others to supply test cases.
# Collect the tests, run them.
# Update the swiki table regarding the status of the fix
# Feed the fix into the update stream
!! Get more people involved in the reviewing and testing process.
Assign areas of expertise / responsibility to harvester champions.
Why not have 25 harvesters with 5 harvester champions? Newbies are
welcome
as well - they know how to press the 'run all tests button'.
!! Get more applications / packages removed from the core
Get more applications / packages removed from the core from the core so
that the number of fixes diminihes.
Actively seek stewards for the various packages.
Question: Where is the link to the current harvesting infrastructure?
-------
Please ask questions if you think something is written in a style not
easy to understand or is just plain wrong.
At the moment I think it is the best to update the swiki page and send a
copy to this list as this is a very important matter. It actually deals
with the way the Squeak community communicates regarding bugs, fixing
them, assessing the quality of the fixes and finally including them.
I think that involvement of more people is needed. Doing this fixing
process can't be the duty of just 5 volunteers!
-- Hannes Hirzel
More information about the Squeak-dev
mailing list
|