This is an updated version of the GiST patch I posted a few months ago.
Attached is a revised version. This update fixes some newly-added bugs in mark and restore (I forgot to save and restore the current buffer), and replaces two ReleaseBuffer() + ReadBuffer() pairs with ReleaseAndReadBuffer(). (Is this still worth doing? It seems it no longer saves a lock acquire/release, but perhaps it will again be a win in some future version of the bufmgr...)
BTW, this idiom occurs a few times:
if (BufferIsValid(buf))
{
ReleaseBuffer(buf);
buf = InvalidBuffer;
}it would be nice to make this more concise. Perhaps:
InvalidateBuffer(&buf);
although that doesn't make the modification of `buf' obvious. An alternative would be to have ReleaseBuffer() always return InvalidBuffer, so:
if (BufferIsValid(buf))
buf = ReleaseBuffer(buf);Any thoughts on this? (I'm inclined to prefer InvalidateBuffer().)
-Neil
new_gist_patch-4.patch.gz
Description: application/tgz
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
