Research talk:Daily unique page creators

Work log

archive
Tuesday, September 9th -- Reverse engineering page creation from the revision table

Tuesday, September 9th

So, the only time you should ever get a rev_parent_id = 0 should be during page creation. I wanted to quickly check this hypothesis with a query to the revision table. The following query is based on a random sample of pages.

> SELECT page_id, COUNT(*) FROM (select page_id FROM page ORDER BY RAND() LIMIT 30) as page_sample INNER JOIN revision ON page_id = rev_page WHERE rev_parent_id = 0 GROUP BY 1;
+----------+----------+
| page_id  | COUNT(*) |
+----------+----------+
|  4092412 |        1 |
|  4745958 |        1 |
|  6663197 |        1 |
|  6918098 |        1 |
| 10545411 |        1 |
| 15126121 |        1 |
| 17933077 |        1 |
| 21001278 |        1 |
| 21261177 |        1 |
| 21884138 |        1 |
| 22841278 |        1 |
| 22973985 |        1 |
| 23918337 |        1 |
| 27643800 |        1 |
| 30124110 |        1 |
| 32762258 |        1 |
| 33503979 |        1 |
| 33766304 |        1 |
| 34316332 |        1 |
| 35341283 |        1 |
| 35649955 |        1 |
| 36075561 |        1 |
| 36546809 |        1 |
| 38088059 |        1 |
| 38752153 |        1 |
| 41010195 |        1 |
| 41538770 |        1 |
| 41703527 |        1 |
| 41707634 |        1 |
| 42775010 |        1 |
+----------+----------+
30 rows in set (40.75 sec)

Looks like this is going to work. --Halfak (WMF) (talk) 21:54, 9 September 2014 (UTC)