Ha @asmecher, I did have a look.
First I ran:
SELECT (*) FROM authors a LEFT JOIN submissions s ON (a.submission_id = s.submission_id) WHERE s.submission_id IS NULL
And found five instances that I deleted.
Then I ran:
SELECT c.citation_id FROM citations c LEFT JOIN submissions s ON (c.submission_id = s.submission_id) WHERE s.submission_id IS NULL;
That resulted in a whole bunch, most likely the product of a test journal that was delete some time ago.
I did delete the whole bunch number produced by the query.
If I now run both queries I get zero as result. That is a good sign I guess. But the update keep failing because of the database error: Duplicate entry ‘0-1’ for key ‘citations_publication_seq.
Is there any other query that makes sense to get me out of this mess?
By the way, after the failed upgrade the table “citations” in the MySQLi database lists 0 as publication_id for all citations.