I would like to see the benchmarks myself
Not as benchmarks per se, but as regression alerts.
Brian already has tweaked the parser so that the number of parser
collisions are only allowed to go down over time, never up.
I think that a time series of benchmarks should be treated the same
way, losing benchmark metrics between checkins should be potentially
considered a bug.
Such a pity that we cant trust the users and the salesman to not use
them...
On Nov 19, 2008, at 11:34 AM, Sheeri K. Cabral wrote:
On 11/19/08, Monty Taylor <[EMAIL PROTECTED]> wrote:
Brian Aker wrote:
> Hi!
>
> I would prefer folks to not publish benchmarks at this time. I
already
> get too many users who wish to make use of drizzle, and benchmarks
would
> just encourage that. Tarballs will come soon.
I agree, of course.
In direct response to Sheeri though... I know a guy who has a whole
big
dbt2-running set up in his basement who runs lots of benchmarks
against
MySQL who is waiting for the go ahead to start benchmarking drizzle...
Nothing wrong with benchmarking now, just don't make those results
known. It might actually be good to get the baselines while
development is happening anyway...
But yeah, I completely agree -- even if there wasn't a resource
issue, it's kinda silly to publish benchmarks when there isn't even
a release version yet. that's the exact kind of "hype you up" crap
that I've seen very often, and it's best not to do much of that,
lest you disappoint folks who expect the world.
-Sheeri
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
--
Mark Atwood <http://mark.atwood.name>
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
