MARK CALLAGHAN wrote:
> On Thu, Mar 5, 2009 at 8:58 AM, Ingo Strüwing <Ingo.Struewing@stripped>
> wrote:
>> Hi Mark,
>>
>> MARK CALLAGHAN, 05.03.2009 17:39:
>>
>>> On Thu, Mar 5, 2009 at 8:25 AM, Ingo Strüwing
> <Ingo.Struewing@stripped> wrote:
>> ...
>>>> Having someone to drive updates of the common style guideline is a good
>>>> idea. I also agree that the changes should be somewhat democratic
>>>> legitimated (e.g. by surveys).
>>> It isn't democratic for me and I contribute a lot
>>
>> Hm. There must be a misunderstanding. My assumption may be unrealistic,
>> but why do you exclude that community developers can participate in a
>> survey or some similar means to collect opinions?
>
> I am not excluding the community. Sun/MySQL is and as a result much of
> the energy from the community has moved over to Drizzle.
>
> I don't mind surveys but I am not sure how many decisions will be
> reached if they are to be made by direct participation rather than a
> representative democracy.
From a Drizzle perspective (may or may not work for MySQL, but figure
I'd speak up) decisions are made via the mailing list consensus.
Usually, it's just a proposal, then +1, -1s, and then someone eventually
sums up the proposal with a decision saying "ok, it looks like folks
prefer XYZ...".
This has worked pretty well so far, but I think the key has been being
as open to change as possible, and trying to understand differences in
opinion instead of being really rigid about things. Generally, if
someone chimes in with a -5 or -10, it means quite a bit more than a few
+1s, especially if the opponent explains their objections.
Overall, it's very helpful to have a singular style, but we all know
MySQL's code base is developed by lots of people and therefore it's
tough to be consistent all the time. Once a code style is decided on,
the bigger problem is to be consistent in code reviews, and make style a
reason for rejection of a patch. The rejection doesn't need to be
harsh, just something like "Looks good, but please fix up some stylistic
stuff to correspond to our coding guide, which you can find here (link..)"
In my experience, surveys don't tend to produce very good results
because they typically don't involve people actually *commenting* and
writing about their preferences, so you don't get a feel for the
opinions of the contributors...you just see percentages. That's why I
prefer the mailing list decisions.
Just my 2 cents.
Cheers!
Jay
