Not sure if this is related to T205594 but in my work on T213597 I found that page_creation_timestamp in wmf.mediawiki_history is not always correct:
USE wmf; SELECT wiki_db, IF(event_timestamp = page_creation_timestamp, 'matches', 'does not match') AS initial_rev_page_creation_ts, COUNT(1) AS n_pages FROM mediawiki_history WHERE snapshot = '2018-12' AND wiki_db IN('enwiki', 'commonswiki') AND event_entity = 'revision' AND revision_parent_id = 0 -- initial rev AND NOT revision_is_deleted GROUP BY wiki_db, IF(event_timestamp = page_creation_timestamp, 'matches', 'does not match');
So looks like it affects ~0.2% of pages on Commons and 1.2% of English Wikipedia pages (0.7% of articles specifically):
wiki_db | does not match | matches | proportion of total |
---|---|---|---|
commonswiki | 134788 | 69274159 | 0.194% |
enwiki | 572631 | 46457894 | 1.218% |
Examples
USE wmf; SELECT page_id, page_title, event_timestamp, page_creation_timestamp FROM mediawiki_history WHERE snapshot = '2018-12' AND wiki_db = 'commonswiki' AND event_entity = 'revision' AND revision_parent_id = 0 -- initial rev AND page_namespace = 6 AND NOT revision_is_deleted AND event_timestamp != page_creation_timestamp LIMIT 100;
page_id | page_title | event_timestamp | page_creation_timestamp | revision history link | first entry in revision history |
---|---|---|---|---|---|
2721713 | U.S._Territorial_Acquisitions.en.alt1.jpg | 2007-09-10 16:27:25.0 | 2012-06-16 12:04:19.0 | revision history | 16:27, 10 September 2007 |
30269399 | UN_General_Assembly_Resolution_66_(1).pdf | 2013-12-21 08:19:34.0 | 2013-12-23 17:36:24.0 | revision history | 08:19, 21 December 2013 |