I've a web application. I'm logging data related to the landing page and subsequent selected hits. create table track_hit ( hitid serial not null, /* pk? I don't mind if serial wrap around pk could be (hitid, tracking_time) */ esid varchar(32), -- related to session tracking_time timestamp not null default now(), -- some other stats -- first_hit boolean not null default false, -- ??? ); I'd like to be sure I just count one "first hit" in a session (same esid) in an efficient way that means I'd like to mark them with a flag and avoid a group by, min subquery cycle when I'm doing reporting. I can't trust the browser and I just want one first hit for each esid, no matter if they have the same tracking_time. Of course the problem is concurrency, but maybe I missed some pg trick that could help me. I'm on 8.3 and no plan to move to 8.4 shortly, so no windowing functions that will make reporting easier/faster. -- Ivan Sergio Borgonovo http://www.webthatworks.it -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
