| From: | Andrew Gierth <andrew(at)tao11(dot)riddles(dot)org(dot)uk> |
|---|---|
| To: | Mark Dilger <hornschnorter(at)gmail(dot)com> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: Hash support for grouping sets |
| Date: | 2017-03-15 21:51:55 |
| Message-ID: | 8760ja43ko.fsf@news-spur.riddles.org.uk |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Another small update to the patch, this time to eliminate any
possibility of integer overflow when handling extremely large estimated
groupings.
Key change:
- k_weights[i] = (int) floor(sz / scale);
+ /*
+ * If sz is enormous, but work_mem (and hence scale) is
+ * small, avoid integer overflow here.
+ */
+ k_weights[i] = (int) Min(floor(sz / scale),
+ k_capacity + 1.0);
--
Andrew (irc:RhodiumToad)
| Attachment | Content-Type | Size |
|---|---|---|
| gshash25.patch | text/x-patch | 149.5 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Andres Freund | 2017-03-15 22:06:10 | Re: WIP: Faster Expression Processing v4 |
| Previous Message | Joe Conway | 2017-03-15 21:46:23 | Re: scram and \password |