On Dec 22, 2007 11:57 PM, Brett Porter <[EMAIL PROTECTED]> wrote:
>
> On 23/12/2007, at 9:35 AM, Jerome Lacoste wrote:
>
> > I'll take the example of the webstart plugin. It's a 2 years old
> > plugin, and although the next release is supposed to be 1.0-alpha-2, I
> > consider it to be good enough for production.
> > The reason it is not yet 1.0 is that I want to keep the ability to
> > make changes to the configuration before 1.0.
>
> Then I think you need to ask - is it really good enough for
> production? A large number of users will simply not use it if they are
> concerned it's subject to change
I understand the thinking. I've just been used to use projects that
haven't reached the 1.0 state for a long time, so it's not a problem
for me.
> Besides, if it's two years old - isn't that configuration pretty
> stable? If not, what is it going to take to make it stable?
Me to get the time to implement a couple of missing features (won't go
into the details as they are pretty plugin specific) :)
> I would
> suggest you call it 1.0 now, meet the other criteria (though in
> webstarts case it probably already does) and *if* you need to make
> that breaking change in the future, call it 2.0 and tell everyone how
> to migrate to that.
>
> The Maven ecosystem needs a culture change from everything being an
> alpha. It's really become an excuse to not finish anything, and it has
> caused a raft of problems: it either doesn't get adopted because it's
> pre-release, or it does and then the developer is forced to not make
> breaking changes (despite it being "alpha") or make them and annoy the
> current users.
It hasn't been a big problem to me. :)
I am more concerned with the complexity sometimes induced by maven,
the breakage in the main plugins/maven itself and the bad POMs one
find on the central repositories.
> This is certainly not isolated to Mojo, but I think it's a place we
> can take simple steps to make a change - it won't affect current
> users, and it can immediately set the tone going forward.
OK
> > I also use some plugins like the shitty one which are not 1.0 but that
> > I would gladly consider using on any project.
>
> Yeah, I'm not saying they shouldn't be used, just that some clearer
> expectations are set.
>
> And frankly, I hope it will push plugins to try and aim for simpler,
> more complete first releases, and to continue that trend with more
> frequent, complete releases after that.
>
> Does all that make more sense?
Yes.
Do we need to do everything at once ? can we start with 1,2 and 3 then
look at 4 afterwards ?
Note, I don't want my -1 to be binding. I.e. if the community decides
its the right way to go, I will comply. mojo is for me a community and
a facility to host some of my plugins.
> > Misc:
> > +1 on the Jira project per project (there should be a checklist of
> > things to do when releasing a project ouf ot the sandbox)
>
> Cool.
>
> >
> > -1 on the test coverage as a criteria
>
> Yeah, I think that would be premature. :)
premature and not necessarily a good goal. E.g.
* one can have a high code coverage but a bad quality (let me test
those getters/setters to increase my rate)
* code coverage tends to be used on unit tests, and I prefer
integration tests to validate the functionality
BTW, are there any statistics of plugin downloads ? That could be an
interesting way to measure popularity and stability.
Cheers,
Jerome
---------------------------------------------------------------------
To unsubscribe from this list please visit:
http://xircles.codehaus.org/manage_email
