Re: SSI using rw-conflict lists

Heikki Linnakangas <heikki.linnakangas@enterprisedb.com> wrote:
> Maybe invent a short name for PredLockTranList and/or the fields?
Done:
http://git.postgresql.org/gitweb?p=users/kgrittn/postgres.git;a=commitdiff;h=44c8f221100b87fc7c23425f53f2fd38d735b7d2
> I'd suggest a union field.
Done:
http://git.postgresql.org/gitweb?p=users/kgrittn/postgres.git;a=commitdiff;h=c6194da36186f8b13785cf606cda2a77f4422afc
Thanks for the feedback!
By the way, the serializable branch on my git repo is back to stable
now, if anyone wants to play with it.
Before going on to finish the graceful degradation work per previous
discussions, I'll try integrating the "IDLE IN TRANSACTION
cancellation" from the current CF to provide feedback from a
practical use case.  If Andres's patch works in that context, it
will allow SSI to provide a "guarantee of progress" even in the most
contrived and pessimal workloads.
-Kevin