Re:PostgreSQL anti-patterns: read-modify-write cycles (2014)
by kornish on Wednesday December 21, 2016 @06:01AM
(#13228220)
While time is down, it's the time of a hypothetical user entering the command, not the time at which commands actually affect data in the table. I think that's pretty intuitively clear to the reader, especially since the article is about counterintuitive behavior.
[ Parent
| on HN ]
Re:PostgreSQL anti-patterns: read-modify-write cycles (2014)
by quizotic on Wednesday December 21, 2016 @06:46AM
(#13228399)
The point is that if a user on the right entered the update command, it would NOT return until after the transaction on the left committed, because there's a database lock on the balance when the left transaction did its update.