[Sugar-devel] A Tale of Sugar and Pippy

[James Cameron]

On Tue, Jun 22, 2010 at 03:39:46AM -0400, Michael Stone wrote:
> After playing for some time -- perhaps 10 rounds -- we discovered that
> we had lost track of which ball was currently contested.

Yes, I discovered that also in my testing of the example.

> We sat down to fix the problem. Since no example was available for how
> to set the color of an already constructed ball, I had to go "behind
> the scenes" by grepping the Pippy source code. Then I was able to work
> out exactly what to do by several small experiments with dir() and
> with "raise".

You discovered what I had discovered ... the example depends heavily on
the Physics library bundled with Pippy.  I got lost looking at the
problem and gave up.  But I did almost manage to convert the code to be
screen resolution independent.  See 4a50004 ... the winning round
position check code has a FIXME attached, and I welcome input.

>    -1: I think there's an important missing "stepping stone" here --
>    I'm not convinced that most people would have been able to figure
>    out how to set the ball color from the currently available
>    view-source interface and Pippy training materials.

I agree.

> Here, we reach the end of my tale. You see, my friend and I agreed
> that our desired next step would be to send our change to sugar-devel@
> along with, well, this story. 
> 
>    -1: Unfortunately, there's no obvious way to do this with Sugar and
>        Pippy today. 
> 
> Anyone have thoughts on what "stepping stones" Sugar and Pippy ought
> to provide to make this act of reflection and sharing feel as natural
> as the act of starting Pippy or of making the change that we want to
> describe and to share?

Quandry: we'd want subsequent users of the examples to be challenged by
the same problem, so why would we want to fix it for everybody?  When
editing the examples recently I saw several improvements that I could
make but decided not to make them because I wanted the reader of the
example to make the same mistake as part of their learning.

Sharing in class context ... does this work?  Can the journal entry be
passed around?

For merging the improvements as part of the Infinite_monkey_theorem, we
would need to bring the change back from the user into the development
community, and provide feedback to the user.  We might not be able to
depend on an e-mail path.  So I envisage a small web application on
sugarlabs.org that would provide these features:

1.  submission of example improvements, which are inserted into a branch
in a git repository, where the branch is named for the user,

2.  status view of their submission branch, using a journal entry of a
Browse bookmark,

3.  download of other submission branches by users,

4.  scoring and voting by other users.

Is there a web application that does this kind of thing already?

-- 
James Cameron
http://quozl.linux.org.au/