út 30. 10. 2018 v 22:05 odesílatel Jim Finnerty <jfinnert@amazon.com> napsal:
re: There is also nonstandard behavior if BEFORE triggers modify rows or prevent updates during an update that is caused by a referential action. This can lead to constraint violations or stored data that does not honor the referential constraint.
That supports the position that it was already documented, I'll grant you that. It is clearly a bug if it causes foreign key constraints to be violated, documented or not. Otherwise foreign key constraints mean nothing in PostgreSQL.
I don't think this would be particularly difficult to fix so that it causes the statement to fail when there is a referential action, rather than silently returning and leaving the database in an inconsistent state.
What is the argument in favor of retaining behavior that renders foreign keys meaningless?
What is the process for deciding whether a report represents an actionable bug that ought to be fixed by the community?
You should to send mail to pgsql-hackers mailing list and send proposal to change the behave.
When there are complex bugs, then you should to design a solution. The bug report is nothing when there is not a possible solution.
The design of BEFORE trigger (NULL returning) is often used pattern and there are not any chance so Postgres drops this feature.
Lot of similar issues are solved by documentation bugfix. The risk is better described.
