> + /* I use CMD_UPDATE, because no CMD_MOVE or the like
> + exists, and I would like to provide the same
> + kind of info as CMD_UPDATE */
> + UpdateCommandInfo(CMD_UPDATE, 0, -1*estate->es_processed);
I do not think it is a good idea to return a negative count for a
backwards move; that is too likely to break client code that parses
command result strings and isn't expecting minus signs. The client
should know whether he issued MOVE FORWARD or MOVE BACKWARDS anyway,
so just returning es_processed ought to be sufficient.
Otherwise I think the patch is probably OK.
regards, tom lane
