Kai Blin wrote:
On Saturday 18 April 2009 05:21:20 Dan Kegel wrote:
http://yokozar.org/blog/archives/48 is a fun little look
at using simulation to see how various strategies
might affect Wine development.
Interesting, but largely academic.
Fair enough. The fact that growth in applications working and happy
users is roughly exponential is interesting though, and I think that
reflects reality.
The one that worked out best was to pick some random
user who's almost happy, fix the last few bugs that
are keeping his apps from working, and then once
he's happy, move on to the next such user.
The problem seems to be identifying these people. The model assumes that you
can tell which piece of software almost works, and that you know the almost
happy users. In reality, you only seem to hear from the pretty unhappy users
and the occasional really happy user. Susan Cragin is about the only user I
can think of off the top of my head who's almost happy and could be made
completely happy by fixing all of the remaining bugs in DNS.
Well, she seems like a nice enough person, why not pick her ;)
But, yes, it's not very helpful when identifying which bugs are
affecting a particular app is half the work (and then identifying how
that API is supposed to work is most of what's left). Still, it's nice
to know that, when we do know, it's not a bad stategy to just go ahead
with it rather than work on something else.
Also, reality has us deal with the fact that new applications are added while
we're working on the old ones, and looking at the graphs, we're only going to
make a significant number of users happy when we're about 98% done fixing the
bugs. I realize that it's a bit hard to model "rate of new applications with
new bugs being added", but that's what happens in real life.
I don't think it's too inaccurate if we imagine the start of the model
being today rather than 16 years ago when the project got started. So
that way we don't have to quite worry about the moving target so much.
Thanks,
Scott Ritchie
