On 03/23/2015 01:31 PM, Ted Dunning wrote:
Here's a better not-quite-so-hypothetical example. A project like MyFaces
>has to pass the TCK testing suite provided by Oracle. We would not want
>to allow unrestricted commit access by someone who did not
>understand profoundly and intuitively that the JSF API portion of the
>project has a predefined public API which cannot be changed.
Some projects feel this way. Others have found that review is just as
effective as restricting commit bits tightly. The classic case is
Subversion which has a very high profile (and thus is obliged to have
stable API's). That PMC offers a commit bit to anyone who asks.
People seem to forget that erroneous commits that pass review can simply be
reverted. That is one of the points of using version control.
Well, sure, and as the board, or as comdev, we can say that that's best
practice, and encourage projects to adopt that. But we're not going to
*require* that they do that. Because they might have a different way
that they want to do things, and it's not our job to tell them that
their reasons are invalid.
The ASF exists to facilitate project communities. Not to dictate to them
how they should run their project. Just to make a welcoming place for
them to operate. As such, we keep the MUST rules to a minimum, and the
SHOULD suggestions are where most of our discussion happens.
--
Rich Bowen - rbo...@rcbowen.com - @rbowen
http://apachecon.com/ - @apachecon
