[development] D7 contrib module development

[Ivan Sergio Borgonovo]

On Mon, 09 Mar 2009 17:52:02 +0100
Marcel Partap <mpartap at gmx.net> wrote:

> >     How could it be integrated with other modules, what could be
> >     factored out into 'frameworks' (/library modules) ?
> > It happens naturally.
> Very well. But there's a reason bioengineering is the latest hype:
> it is a lot FASTER and more DIRECTED than the evolutionary process.

Faster and more directed doesn't mean right.
I think that evolution isn't rational... but at least it generally
provide a lot of testing ;)
We can try to be rational and let evolution do the "testing".

> > What I am saying it happens naturally as people realize that
> > they are evolving their pet project into an API/Framework over
> > time.
> At our university we are taught: the more brainwork you invest
> early on in the development process, the less headache you'll have
> to cope with.

If you invest it right and you're freeze in time.
I'm not a strong fan of XP but still in its less X incarnation has
its merits.

The problem of low quality modules and duplication could be reduced
to further info on module pages: activity, releases, downloads,
related modules...
These aren't perfect metrics but they can help without anyone
playing God or just get into the way of developers making
requirements.



> > You can't dictate what happens in contrib without stifling the
> > innovation that this ecosystem has.
> Who's talking about dictating.. Everyone can code freely. But for
> code to enter the official repository, quality checks should
> apply. That doesn't stop innovation, does it? Look at the few
> commits to the cck/views branches for example. Why is that kind of
> responsible code committing not scalable to all other modules as
> well?
> 
> > We can nudge people to work together but we can't force them to.
> Well not many people who score goals by hand have made successfull 
> career in football. How comes? Might it be that whoever doesn't 
> respect the rules will not be allowed to play?
> 
> 
> >> I want to have *all* the functionality spread over slightly
> >> different modules with the same purpose, combined into one
> >> flexible solution. And i am quite sure i am not the only one.
> > You are advocating the "one true way".
> Am not. What i am advocating is a process that makes problems that 
> occured in the past less likely to happen.
> 
> 
> 
> > But without letting people
> > experiment
> Why is not letting experimental code into the repository stopping 
> people from trying stuff out? Please explain that to me, i simply 
> don't get that.
> 
> > that one true way could be the one with less features, with
> > a non-committed maintainer, could be buggy, ...etc. You don't
> > know until you put it out there and let the ecosystem decide.
> 
> > Who will decide the one true way? Committee? Hierarchy?
> No - the swarm, that is us. We all work together on finding the
> best way to lay the foundation for the respective functionality.
> Aren't we already doing that with core? Is it impossible to apply
> to non-core modules?
> 
> > What we can do is let things evolve for a while and then
> > naturally a winner or two will emerge from the fray.
> So please tell me, what is the winning change notification
> framework? Notifications, Subscriptions, Comment Notify,
> Notify, ...? Which is the best module to create sophisticated web
> questionnaires? Is it webforms? Advanced Poll? Decisions?
> Do you really want to have the same situation with D7? Not
> terribly user-friendly, is it?
> 
> 
> > As disconcerting this is to
> > some, it is a sure way to have a proven solution for the problem
> > space.
> Is not. It might be sufficient, but it surely isn't the optimum.
> 
> > No. I am saying is that without seeing people contribute for
> > some time you don't know before hand if they will turn to be a
> > drive-by contributor of a so-so project, or the author of the
> > next big hit.
> Why is applying quality checks before applying code to the
> official repository making it harder for people to code,
> experiment, innovate? How does having each submission routinely
> reviewed by the community stop anyone to contribute?
> 
> > All that cannot be seen just from the first contribution
> > adherence to coding standards.
> Whos talking about judging people by their first code? But what is
> a good reason for the first contribution of someone starting to
> learn drupal to be included in the official repository? Wouldn't
> it be much more helpful to receive comments and tips in the
> discussion thread for their patch?
> 
> > Fair point, standards will arise across CMS's too
> Didn't even touch _that_ topic.
> 
> > But within Drupal, I don't see we are the point where we have a
> > central body .
> Well i didn't propose a central committee, but for the whole
> community to pick up quality checking and maintain the modules as
> a huge diverse team. Sounds implausible? Well who's stopping us
> from trying.
> 
> 
> > Later maybe? Perhaps. But not now. When you see a majority of the
> > community cry for that, then it is time to re-evaluate.
> At the time people are crying it usually is a bit late.
> 
> >     rgds marcel.
> 
> 


-- 
Ivan Sergio Borgonovo
http://www.webthatworks.it