Development process analysis

[Gervase Markham]

Guys,

We're doing a lot of thinking about the development process at the 
moment. However, it seems to me that we haven't yet had a chance to 
implement the last round of thinking, and see whether the things we 
wanted to do then actually improve matters. This means that a lot of the 
discussion is merely rehashing the same points from six months ago.

I seem to remember that the consensus was that the following things 
needed to happen:

- More frequent upgrading of b.m.o. for testing purposes
- Shorter development cycles to avoid long stabilisation periods

However, as we haven't yet released 2.18 or 2.20, we haven't had a 
chance to put this into practice!

If, once we branch for 2.20, instead of saying "according to the 
schedule, we are now three quarters of the way through the 2.22 
development cycle; so we only have a month to get patches in", we say 
"OK, the 2.22 development cycle starts now", we can get back on track 
and start trying out some of the last round of process improvements. 
Then, we can see if they actually work! :-)

(I'd argue we actually want to move to nine-month cycles, but that's a 
different discussion.)

Gerv