Re: Problem with recent PostgreSQL relatedpressrelease - Mailing list pgsql-advocacy

Dear colleagues,

On Fri, Jul 13, 2007 at 11:02:40AM -0400, Derek Rodner wrote:

> commercial vendor follows.  After all, we don't want other companies
> hobbling our technology and claiming they are x times faster.  It is one
> of the benefits of being a proprietary product.

This seems to me to be unobjectionable, because the BSD license in
fact does give people the ability to close their product code, and
place such restrictions on it, and that seemed to be part of the idea
behind the BSD license in the first place.  On the other hand,

> Second, let's talk about the "up to 200% faster" claim.  I stand behind
> this statement 100%.  After all, they are my words.  And, remember, as
> Bruce said, it is marketing.

this _is_ objectionable.  To begin with, the glib remark that it's
"marketing" suggests that tossing mud at others is ok, as long as
it's all in the spirit of selling stuff.  Second, and somewhat more
importantly, I think, the claim is an apples-to-oranges comparison
that the community is _not allowed_ to rebut, because EnterpriseDB
isn't willing to let us.  It's reasonable to expect such behaviour
from other database companies, but it seems a little like eating
your young in this case.

Note that I actually wouldn't have any complaint about the "200%
faster" claim if the claim simply said _why_: "EnterpriseDB's
sophisticated Dynatune tool automatically provides performance gains
of up to 200% over untuned PostgreSQL," (or something similar)
wouldn't cause me to bat an eye.

Quite apart from issues in the community, though, I wonder whether
this angle isn't subtly undermining EnterpriseDB's own arguments.  If
PostgreSQL is so great, how come people are able to get 200%
increases in performance out of mostly the same code without much
work?  Why isn't the super-excellent PostgreSQL code already doing
that?  Oh, and some of these things are new features?  Ah, so it's
_not_ just PostgreSQL, tested by the community &c. &c.?  I don't
think you can make the argument work in EnterprisDB's favour on both
sides.

I am very far from being someone who objects to closed bits of code,
to people doing what they want, &c.  I pushed hard at Afilias to get
Slony (and some other code) released under the BSD license partly
because I think it's good for the community, even if part of the
community goes away and makes changes in secret (before or without
sharing those changes over the long term).  But I don't think it's
cricket to be slyly suggesting that the basic code is so much worse
that the magic of EnterpriseDB's contribution solves everything, and
then tell the community they can't test it themselves and publish the
results.  It might be permitted by the license, but it isn't part of
any community behaviour I can think well of.

Best regards,
A

--
Andrew Sullivan  | ajs@crankycanuck.ca
This work was visionary and imaginative, and goes to show that visionary
and imaginative work need not end up well.
        --Dennis Ritchie