1 Jul 2005 01:02
Re: Software development without pairing?
<bena <at> xenon.triode.net.au>
2005-06-30 23:02:31 GMT
2005-06-30 23:02:31 GMT
I don't know that I said "measure everything you can". I hope I said
"look at". That is, look for sources of information that gives you
visibility into whatever it is that troubles you.
I use measurements when I have a problem and my usual experience is that
finding the right measurement exposes the root cause and allows it to be
fixed, making the measurement useless.
For example, I had $THING which would sometimes run and sometimes not. I
set up a script to run $THING regularly, and complain loudly if it didn't
work. I soon saw that doing $OTHER_THING was always followed by screams
of complaint from the script. Although $OTHER_THING worked, it was
corrupting some data $THING depended on. $OTHER_THING was changed, the
problems with $THING went away and the script had no further value.
> This means deciding what the problems are
> first, then picking metrics that help us understand those problems, and
> only *then* measuring -- not the other way round.
I'd suggest both. There's not much worse than a person with a solution
desperately looking for a problem to apply it to. But sometimes, when you
hear of a solution, you realise it solves a problem you didn't realise you
had.
Regards, Ben
William E Caputo <wecaputo <at> thoughtworks.com>
Sent by: extremeprogramming <at> yahoogroups.com
30/06/2005 00:15
(Continue reading)
RSS Feed