On Thu, Jan 23, 2020 at 5:31 PM Peter Geoghegan <pg@bowt.ie> wrote:
> In summary: I suppose that we can also solve "the cousin problem"
> quite easily, but only for rightmost cousins within a subtree --
> leftmost cousins might be too messy to verify for it to be worth it.
> We don't want to have to jump two or three levels up within
> bt_downlink_connectivity_check() just for leftmost cousin pages. But
> maybe you're feeling ambitious! What do you think?
I suppose the alternative is to get the high key of the parent's left
sibling, rather than going to the parent's parent (i.e. our
grandparent). That would probably be the best way to get a separator
key to compare against the high key in the leftmost cousin page of a
subtree, if in fact we wanted to *fully* solve the "cousin problem".
Goetz Graefe recommends keeping both a low key and a high key in every
page for verification purposes. We don't actually have a low key (we
only have a truncated negative infinity item), but this approach isn't
that far off having a low key.
--
Peter Geoghegan