[squeak-dev] Compromise

Given that not only are Andreas and I talking at cross purposes, but
also our processes are virtually orthogonal, I have concocted a compromise.

X) My 3.11 proposal is about harvesting contributions that are complete.
Y) Andreas' proposal is about getting people motivated to begin
contributing.

Due to the fact that I need to eat, and pay my bills, I am unable to do
anything for at least 12 weeks, I am sorry but that is how the cookie
has crumbled this week.

1.

I propose that Ken organise the Mantis crew (you know who you are) to
pick the top 100/200 or 300 bugs, and to test them in latest
"3.10.2-build" available from

http://ftp.squeak.org/3.11/Squeak3.10.2-build/090628-1523/ 

Mark those tests you are most confident in as "Resolved" - "fixed in
3.11", that is all you need to do to harvest a fix.

2.

Andreas to review and edit the image re-organisation tasks as defined in...

Installer install: 'Packages'.
Installer ss project: 'Bob'; install: 'Tasks-Squeak310'.

Squeak310PlusOne-#taskReorganizeCategories.
Squeak310PlusOne-#taskReorganizePackages.

Define your own ideal image organisation, hopefully following the model
provided which puts tests in context next to the category they are
testing (for a much tidier image)

Add to the above additional code to split up System and Network into
smaller pieces that can be farmed out. (Please someone find a new
HTTPClient/HTTPSocket)

Reorganise "3.10.2-build" using the above, and save to the "trunk"
repository, so that new contributions will use the new "designed"
organisation of packages.

3.

Andreas, Eliot, Igor, Matthew et al.

Pick/Instigate 2-10 projects to carry out in the newly organised trunk
(or elsewhere as appropriate), each aiming to deliver a completed "grand
refactoring".

One of these could/should be the re-licence, and a second should try and
fix our changes file limit problems, a third should be integration of
closures.

4.

Göran and Igor

Hammer away on DeltaStreams so that we have a cool way to share and load
these "grand refactorings".

5.

Keith

On my return, I will import all of Mantis into MC, so that the chosen
list of bugs is remembered (I will put the code to do this is into
squeaksource/Bob when it lets me).

I will then configure Bob to build the 3.11 release alpha/beta with the
300 best fixes, and any completed grand refactorings. Which can be
applied before or after the fixes as desired.

How about it?

Keith

On Tue, Jul 07, 2009 at 03:14:06AM +0100, Keith Hodges wrote:
Keith, I quite literally just ran out of disk space on my laptop PC
while trying to download the 13MB image that is apparently required to
bootstrap the task list. No kidding. I trust that the actual list of
incomplete tasks requires something less that 13MB to describe; if so
a brief summary would be helpful.

Dave

David T. Lewis wrote: The 3.10.2-build 13Mb image, is 3.10.2 with the latest tools MC etc and
a few fixes. Sake/Tasks/Packages are all easily unloaded later on.

I suggest that you point your MC Browser at the Bob repository and have
a look around. The tasks are present, but need updating to reflect the
latest Sake/Package API

Keith

On Tue, Jul 07, 2009 at 04:57:44AM +0100, Keith Hodges wrote:
I will do that, thanks.

Dave

On Tue, 2009-07-07 at 03:14 +0100, Keith Hodges wrote:
> I propose that Ken organise the Mantis crew (you know who you are) to
> pick the top 100/200 or 300 bugs, and to test them in latest
> "3.10.2-build" available from
>
> http://ftp.squeak.org/3.11/Squeak3.10.2-build/090628-1523/ 
>
> Mark those tests you are most confident in as "Resolved" - "fixed in
> 3.11", that is all you need to do to harvest a fix.

I'm a little bit confused here.  Shouldn't we mark the issues as 'test
for 3.11' so that Bob gets it's pass on the issues before they are in
fact considered good for harvesting?  Perhaps I'm missing the point but
I'm assuming this is you making clear and wishing to 'practice' one
phase of the 3.11 plan which involves a little human pre-checking of
issues before Bob is let loose on them.  Is that correct?

Ken

On Tue, 2009-07-07 at 18:17 +0100, Keith Hodges wrote: OK, I'm going to have to think about this a bit before committing myself
to it.

Ken

Hi Keith -

I really appreciate your attempt to compromise. I appreciate even more
that you are acknowledging that the processes are orthogonal and not
contradictory.

My biggest concern with the proposal is that you seem to be committing
other people's time over a period of time that you yourself will be
unavailable. Considering that the people who you've signed up may have
other plans and that you are not around for any guidance, how well do
you think this will work?

Secondly, you are adding several new dependencies in the process of
getting to a result, including "2-10 grand refactorings" plus
DeltaStreams. That is unrealistic. Nobody can commit to grand
refactorings that haven't even been proposed.

Can't we just choose the available fixes from Mantis *now* and apply
them, and build a 3.11 release candidate *today*? This wouldn't require
committing time from other people, it wouldn't put additional
dependencies in the way; all it requires is to sit down and actually run
those scripts and produce an artifact from it.

Cheers,
   - Andreas


Keith Hodges wrote:

Andreas Raab wrote:
> Hi Keith -
>
> I really appreciate your attempt to compromise. I appreciate even more
> that you are acknowledging that the processes are orthogonal and not
> contradictory.
>
> My biggest concern with the proposal is that you seem to be committing
> other people's time over a period of time that you yourself will be
> unavailable.
If things had worked out differently then I might have been more
available. Having recent somewhat unpredictable events has adversely
effected my deadlines, and I need to make a concerted effort to get back
on track otherwise I am really scuppered once and for all.

I am using Bob for a part of a commercial project so I am still
contributing but under the guise of real work.
> Considering that the people who you've signed up may have other plans
> and that you are not around for any guidance, how well do you think
> this will work?
I will be available on irc, and skype for brief chats.
> Secondly, you are adding several new dependencies in the process of
> getting to a result, including "2-10 grand refactorings" plus
> DeltaStreams. That is unrealistic. Nobody can commit to grand
> refactorings that haven't even been proposed.
My aim here is to encourage people who are motivated to work on things
anyway, to be able to work on things that are a) not on a critical path
for 3.11 and b) will be most useful, in such a way as their efforts will
not work at cross purposes to anything that is on the critical path.

Your concern that I am adding dependencies is not the case. We are not
dependant upon these projects for 3.11, however we are dependent upon
these if we want to have anything but fixes to include for 3.12. I would
like to see atomic loading in there though, and its not like we are
short of new/better stuff in 3.11-build (the version of 3.11 for
building images, as opposed to 3.10-build which is our starting point).
> Can't we just choose the available fixes from Mantis *now* and apply
> them, and build a 3.11 release candidate *today*?
Yes we can, and as I explained it would take very few lines of code in
bob to do just that. But I envisaged that in the time I take to get
around to it, lots of people would have started harvesting fixes and
tests into trunk, and as a result changing the status levels in mantis
and generally turning what is already a confused mess into something
even worse.

So I figured that if the best coding folks could be motivated to do real
useful projects, then they wouldn't start spending their time
harvesting. And if the bug busting, mantis crew could be motivated to
test and consolidate what we have already, they too would be doing what
they know and love(?) best, and would be making things better for the
scripts when they are run.
> This wouldn't require committing time from other people, it wouldn't
> put additional dependencies in the way; all it requires is to sit down
> and actually run those scripts and produce an artifact from it.
And that is what I don't have time to do right now even though it is
probably trivial.

The structure that I have in mind is

Squeak3.10.2-build --> BobBuild310Fixes --> Squeak2.10.2-fixes

#description ^ 'Squeak3.10.2-fixes is a build including all the fixes
that have passed testing for inclusion
into the next squeak release. These fixes are currently marked "Fixed In
3.11" in mantis.

This image includes all of the fixes that 3.10 users might reliably
apply to their images,
and a selection of these fixes may be chosen to form 3.10.3'

Squeak3.10.2-build --> BobBuild310NewFix --> Squeak2.10.2-newfix

#description ^ 'When a fix has its status changed on mantis, this build
action loads that fix into the base image and runs the tests'.

Squeak2.10.2-fixes  --> Squeak310PlusOne buildTestCandidate -->
Squeak3.11-stable-test-candidate

Squeak2.10.2-fixes  --> Load Testing fixes -> Squeak310PlusOne
buildTestCandidate --> Squeak3.11-unstable-test-candidate

regards

Keith