Robert,
> I don't know if they have ever done that before, but other companies
> have done things like that. Had we contacted them privately it would
> have given them the chance to address this case, even if only off the
> record.
The problem is that, unless you personally have a friend on IBM's legal staff,
there was *no way* for us to contact IBM lawyers quietly, off the record.
Any contact we made would be official and start the IBM legal machinery
moving.
And, let me remind you, that while IBM's senior management has been a great
friend to Open Source and our project, their DB2 divsion has been allowed to
conduct a FUD campaign against PostgreSQL in an effort to encourage
migrations. If we inquired at the wrong office based on our lack of
information about IBM internal politics, the consequences could have been
pretty bad.
At the time Tom created the 2Q code, we didn't know when the IBM patent would
be granted. It could have been any day, or even already granted and the
public records' hadn't caught up. Without IBM contact, we would still have
a "grace period" until IBM discovered our use of ARC and sent us a
license-or-quit letter, which they would be unlikely to do quickly (if at
all). If we contacted them and they decided *not* to license us the patent,
or to license it under the GPL, then they would be bound to serve us
*immediately* on patent grant. And then we would be scrambling to replace
ARC and get our users to upgrade in a few days.
Core spent several weeks discussing this and obtained legal advice (not named
here because it was not formally retained) who agreed that replacing the
algorithm was the only "risk-free" course.
--
Josh Berkus
Aglio Database Solutions
San Francisco
