On Tue, 2008-10-21 at 14:11 +0400, Teodor Sigaev wrote:
> > I guess we'd use the same technique for GIN. ginInsertValue() ??
> > Hmm, you release the lock at line 412, ginbtree.c before you get the
> > parent lock at line 428. That seems different to the L&Y interactions.
> > Am I looking in the wrong place?
>
> at line 412 new page (right page) is unlocked, old page (left one) is unlocked
> later - at line 448, after parent page is locked.
Thanks for checking.
-- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support