I like the idea, however I can see this as being too much work.  As someone
faced (admittedly doing it the first time) with releasing some plugins, I
feel like this is too much.

What is really needed is more extensive unit testing of plugins.  For
instance, with the CruiseControl plugin, I have added unit tests that verify
that the SCM plugin works the way I expect it to work.  This has helped me
catch issues when SCM has changed.   More unit tests will help the stability
of plugins.  Additionally, I have my email set up to flag me when a CVS
commit message is about a specific plugin that I'm interested in.

The idea of a release period is nice, but with the amount of manpower
available, I'm just happy to see a plugin released at all.  I called a vote
to get my plugins released three days ago, and haven't (for various reasons)
managed it yet..  So I don't see the level of organization arising to get
every plugin to hit a specific period.

On a related note, I am assuming you are hoping to reduce churn in your
plugins?  I have been working on a plugin that checks for newer versions of
your current plugins and/or project dependencies.  The idea being that you
can quickly pull all the update information versus manually checking it.  I
can share this with you if you like.  It's still a work in progress however.

Eric

> -----Original Message-----
> From: Paul Spencer [mailto:[EMAIL PROTECTED]
> Sent: Wednesday, August 11, 2004 10:25 AM
> To: Maven Users List
> Subject: Plugin release procedure, was Re: Dashboard plugin release
> status?
>
>
> In an effort to balance "access to updated plugins", "quality of the
> released plugins", "user expectations", and "developer resources", I
> purpose the following procedure whenever a plugin has been updated.
>
> 1) Deploy a snapshot of the plugin.  This allows the community access to
> the updated plugin and provide feedback prior to an actual release.
>
> 2) Update the plugin's website, i.e. site:deploy.  This will notify the
> community that the plugin has been updated.  Additionally posting a
> "Snapshot Announcement" to the user mail list would be an nice
> enhancement to this procedure.
>
> 3) Assuming the plugin works, release the plugin during the next
> "Release Period".  This will ensure the quality of all plugins by
> allowing the community time to provide feedback on updated plugins while
> minimizing the amount of time spent on integrated testing.
>
> "Snapshot Announcement" - An announcement, similar to the release
> announcement, but for a snapshot.  The announcement should include:
> o List of changes
> o Download instructions
> o Anticipated release date/period
>
> "Release Period" - A set period, i.e. the first week of the month, when
> all unrelease and working plugins are release.  During this period:
> o Each unrelease plugin must pass all of it's testing.
> o All plugins must be retested, i.e. integrated testing.  This is to
>    verify dependent plugins work.  As an example the SCM plugin depends
>    on the CHANGES plugin, so changes to the CHANGES plugin will affect
>    the SCM plugin.
>
> Paul Spencer
>
>
> Brett Porter wrote:
>
> > We weren't talking about bugs. We were talking about new features.
> >
> >
> >>So you'd rather not release early and often? That has nothing to do
> >>with fixing bugs.
> >
> >
> > But it has everything to do with not introducing new ones.
> >
> >
> >>What if there are bugs fixed in the CVS version, critical ones even. I
> >>don't see why they should wait for all outstanding bugs to be fixed
> >>before releasing.
> >
> >
> > That's why we've started with the bugfix releases. Critical, or a set
> > of minor bugs all fixed. But lets not release after every new feature
> > or whenever some code gets reformatted or some documentation gets
> > added (since that can go to the website earlier).
> >
> > I probably got my wires crossed, but the point I wanted to make was
> > that I think we should be fixing bugs before adding features, and
> > doing more testing before officially releasing plugins after features
> > are added.
> >
> > - Brett
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to