On Thu, 25 Jul 2002 14:40:09 +0100
Matthew Toseland <[EMAIL PROTECTED]> wrote:

> On Wed, Jul 24, 2002 at 10:32:56PM -0500, Ed Onken wrote:
> > At 09:21 PM 07/24/2002 -0500, Ed Onken wrote:
> > 
> > As far as my below fix goes, I don't whether or not we want to fire
> > a CollisionEvent or not when a CHK has a key collision.  I don't
> > know if it would confuse clients or not to have a insert succeed
> > even when a key collision occurs.
> I vote that we ignore CHK collisions at the client end, and just
> document it somewhere.

I do plenty of KeyCollision ignoring for CHKs already. But yes, it is
rather unintuitive, and somewhat undocumented, that a KeyCollision for
a CHK is Really Not A Bad Thing, but for most other types it is
definitely a Bad Thing.

No chance on getting the message changed this late in the game, eh? :)

--hobbs

Attachment: msg03519/pgp00000.pgp
Description: PGP signature

Reply via email to