[haiku-commits] Re: r40520 - in haiku/trunk/src: add-ons/kernel/drivers/graphics/radeon_hd tests/kits/net/preflet/InterfacesAddOn


> However I wonder whether it was a bit more "direct" than usual this time

> due to the very quick way that Alex was granted commit access - there 
> was no official vote, and only replies from Clemens, Stephan, Jerome and

> Philippe before the change was made (Matt did wait 2 days after the 
> final reply before making the change though). There was also no "I'll 
> accept in two days unless there's any objections" note.

We decided to remove "votes" because a single '-1' is usually enough to
abort the process. So it's currently 'speak now or shut up forever'.
I don't think waiting more is going to be useful ?

Anyway, each of us developpers went through build breakage and commit
style police stage at the beginning. It's part of the learning process, and
yes, it is more important than actual functionality.

http://thenextweb.com/shareables/files/2011/01/65987_700b.jpg

The network preflet is the perfect example, with 3 different incarnations,
the 2 first being abandonned because they are not modular enough. Good
design takes longer, but gives better long-term results. This is an
opensource project, with people coming and leaving as free time allows, and
the consistent style really helps with reading any part of the code. Sure,
it may seem boring to argue about the number of blank lines between the
licence comment and the header guard, but the only way out is improving the
"checkstyle" scripts, so devs don't have to do it by hand anymore. Is
getting your commit rejected by a robot less demotivating ?

-- 
Adrien.

Other related posts: