Forum OpenACS Q&A: Re: RFC: OpenACS Governance

Collapse
Posted by Don Baccus on
Peter, you're going to be stuck being on the core team, I'm afraid, why do you think I asked you to start this discussion during our meeting in Copenhagen? :)

As far as commit rights go ... I am going to urge us to continue to be somewhat cautious until we have more people reviewing commits (Jeff eyeballs most of them today, I eyeball a bunch of them, too, does anyone else at the moment?)

Also I want to be somewhat cautious until we have proven ourselves to be able to better test our stuff before release.  On that note I think I'd support giving *anyone* commit rights to a branch of the tree labelled "regression tests" ...

I'd like to point out that we just released 4.6.2 which has one nearly fatal bug in form handling introduced recently by one of our senior hackers (no need to name names since it could've been any of us.)  Empty but required form fields are no longer flagged, as was discovered two days ago (and fixed and will end up in the 4.6.3 release which is forthcoming.)

Rather than point fingers at the person who introduced the bug, I'd like to use this as an example of how pitiful our testing efforts still are.  Much better than they were before 4.5 (and I'd argue better than aD's on average) but no where near good enough.

We've had one case of a committer walking over other people's code to the extent that I was asked to remove their commit rights.  That was early on in the project.  The problem wasn't so much the code they contributed but rather their lack of respect for other people on the project, which was made clear by his refusal to communicate to package porters before changing the code being worked on silently, behind their back if you will.

I don't expect problems of this sort to crop up often so I don't worry about it.

But ... without a better testing process and without more people actively reviewing committed code, I feel more comfortable with newcomers going through the patch submisison process.  This *forces* review before they're applied, that's the main reason I favor it.

On the other hand ... two of the reasons we .LRN TAB folk decided we wanted to push forward on OpenACS governance was to expand the core team in a way that is explicit and recognizable by the community (and doing so should help with the "we need more eyes reviewing commits" problem), and to provide a framework for the kind of mentoring of newcomers mentioned above.

With that framework in place I'd feel a lot more comfortable  dishing out commit rights more freely.

Of course ... we have dished out commit rights quite freely on slices of the tree, and there are a considerable number of people who currently have full commit rights to the tree.

So I don't think we're talking about a black-vs-white comparision, here, simply shades of gray ...