Peter Donald <peter@apache.org> writes:
> On Wed, 6 Nov 2002 14:05, B. W. Fitzpatrick wrote:
> > > I find code ownership a problem that can and must be prevented and
> > > resolved in the community. A trick that seasoned committers do on new
> > > committers is to change their first commits and work on them, to show
> > > that the code is of everyone. If they complain, it's time for a nice and
> > > bold explanation.
> > > From my experience on this, it's not something one forgets easily ;-)
> >
> > Are you serious? Quite frankly, I find that behavior reprehensible--It
> > reeks of strange fraternity initiation rites.
> >
> > If I write, test, and commit a piece of good solid code
>
> If you do that then it would be unlikely to be changed. I think I know what
> Nicola is talking about and I think it is great :) You may also notice that
> the project who have the bigger healthier communities (at least in
> jakarta/xml land) tend to do this all the time.
>
> For example, someone submits some code that doesn't follow various convention
> s
> that have been established in the project. Do you tell the contributor -
> sorry can't take that till you fix it? No. Usually what happens is that you
> commit the code. Then you go through and fix up style/semantic/logical
> violations. As the commit messages go past the end user sees the corrections.
> Next time they are more likely to work the way the project operates.
This is very different than what I understood from Ken's post. I
agree that consistent style is very important--I thought he was
referring to code content. The above scenario is totally
understandable.
> If there are massive fixes required the user will generally see the patch
> rejected with recomendations for a fix but usually it is better to commit and
>
> teach by example IMHO.
>
> > and someone
> > else goes pissing in it just to leave their scent and to show me that
> > I don't 'own' the code, I am *not* going to be amused by it.
>
> Not many people are - they soon learn or leave. Usually a good test of
> maturity ;) If someone has a fit because their curly bracket got moved then
> do you really think they get collaborative development?
Right. As I said above, this makes sense.
In Subversion, we take a slightly different approach on this. When
people send in a patch that doesn't follow style guidelines or include
a log message, we point out the issues ask them to resubmit an amended
patch. After a developer has submitted a number of 'correct' patches,
we will then usually offer commit access.
So it looks like we agree after all. :)
-Fitz
|