[Sugar-devel] [Dextrose] Pending patchs for Paint

[James Cameron]

(composite reply to several postings in thread)

On Fri, Oct 22, 2010 at 03:16:49PM +0100, Gary Martin wrote:
> On 20 Oct 2010, at 21:56, James Cameron wrote:
> > On Wed, Oct 20, 2010 at 04:56:41PM +0100, Gary Martin wrote:
> > > 	http://git.sugarlabs.org/projects/paint/repos/mainline
> > > 
> > > I noticed that the activity.info file is still for version 27. As
> > > Paint-28 has been released I'd expect to mainline to at least be
> > > up to date with those commits. Maybe the rest of the source is
> > > current, and it's just the activity.info file change that was not
> > > pushed?
> > 
> > While it would be nice if every release of every activity has a
> > matching tag and sources in the respective repository, we've had
> > situations where that has not been desirable, such as when a version
> > of Record was released for a deployment via activities.sugarlabs.org
> > but the changes were not accepted into mainline because consensus
> > had not been reached.
> 
> In such cases git.sugarlabs.org provides very easy cloning of any rep
> where deployments can have full control and create, say, a custom
> Record-Peru.

It might not have been easy enough, because it wasn't chosen as the
solution to the issue at the time.

> > A release of an activity is not done using the master repository, it
> > is done using a local clone of the repository.  Therefore during
> > testing and diagnosis I've always presumed that the source code
> > shipped in the .xo file might be different to the tagged source in
> > the master repository.
> 
> Hmm, I'd have to disagree. This seems like a terrible workflow for a
> community. [...]

How would you make activities be released from a master repository?  You
would automate the release mechanism somehow?  You would add one more
person to the release sequence?  Or you would refuse to list activities
on activities.sugarlabs.org unless they match a master repository?

In my opinion, the problem does not happen frequently enough to warrant
this attention, but for diagnosis and testing I must continue to check
that the source matches the master repository.  To not do so in the
absence of strict controls would be negligent.

On Fri, Oct 22, 2010 at 06:45:06PM +0200, Sascha Silbe wrote:
> Excerpts from James Cameron's message of Wed Oct 20 22:56:04 +0200 2010:
> 
> > While it would be nice if every release of every activity has a
> > matching tag and sources in the respective repository, [...]
> 
> I certainly hope that each bundle released by Sugar Labs or associated
> organisations (OLPC, etc). has a corresponding commit and tag in a
> discoverable git repository.

As I said, this wasn't the case.  The frequency of such exceptions is
low though.

> It doesn't need to be the on the master branch or even inside the
> mainline repository. What matters is there is
> a) a corresponding commit in a public git repository
> b) the commit belonging to a particular bundle is discoverable within
> a reasonable amount of time.

It certainly wasn't.

On Sat, Oct 23, 2010 at 10:21:11AM +0200, Sascha Silbe wrote:
> As a developer, it's important to me that the corresponding source
> code for each released object (e.g. activity bundles) is easily
> accessible.

On the other hand, the .xo file does contain the source.

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