Bruce Momjian wrote:
>What exactly was the change? As I remember, it was something we didn't
>think would affect anybody, but it did affect you.
>
Here's the comment from the code (in the CursorResponseMessage class) I
patched ....
"Due to a change in the fontend-backend protocol version 0 2 0 0, I must
check to see if another CursorResponseMessage follows right after the
one that led to my creation. If there is, I take the last one as being
the 'correct' one"
Note: the "I" and "my" in the comment refer to a
CursorResponseMessage instance.
Also, there was a discussion on this list between Tom and I about the
change and the direction it implied. The gist as I recall centered
around the question of whether every side-effect of a query should
return a distinct CursorResponseMessage in addition to the one for the
explicitly invoked query.