Thread
-
11 beta 3 / ROLLBACK TO SAVEPOINT regression in PLPGSQL
David Klika <david.klika@atlas.cz> — 2018-08-23T11:40:38Z
This routine works well on postgres v10: CREATE FUNCTION test_savepoint() RETURNS VOID AS $body$ BEGIN SAVEPOINT my_savepoint; ROLLBACK TO my_savepoint; END $body$ LANGUAGE plpgsql; The same code can not be compiled in postgres v11 beta 3, I got error: [42601] ERROR: syntax error at or near "TO" Position: 117 Best regards David Klika
-
Re: 11 beta 3 / ROLLBACK TO SAVEPOINT regression in PLPGSQL
Sergei Kornilov <sk@zsrv.org> — 2018-08-23T12:41:21Z
Hello > This routine works well on postgres v10: it works? Or is only accepted as input? I got error on function call. I thought plpgsql cannot directly uses savepoints, only implicitly with begin ... exception ... end block ( https://www.postgresql.org/docs/current/static/plpgsql-control-structures.html#PLPGSQL-ERROR-TRAPPING ) Well, you can still create this function with set check_function_bodies to off regards, Sergei
-
Re: 11 beta 3 / ROLLBACK TO SAVEPOINT regression in PLPGSQL
Pavel Stehule <pavel.stehule@gmail.com> — 2018-08-23T13:00:47Z
2018-08-23 14:41 GMT+02:00 Sergei Kornilov <sk@zsrv.org>: > Hello > > > This routine works well on postgres v10: > it works? Or is only accepted as input? I got error on function call. > I thought plpgsql cannot directly uses savepoints, only implicitly with > begin ... exception ... end block ( https://www.postgresql.org/ > docs/current/static/plpgsql-control-structures.html#PLPGSQL-ERROR-TRAPPING > ) > > Well, you can still create this function with set check_function_bodies to > off > This doesn't help, because any function is validated before execution Regards Pavel > > regards, Sergei > >
-
Re: 11 beta 3 / ROLLBACK TO SAVEPOINT regression in PLPGSQL
Pavel Stehule <pavel.stehule@gmail.com> — 2018-08-23T13:04:59Z
2018-08-23 15:00 GMT+02:00 Pavel Stehule <pavel.stehule@gmail.com>: > > > 2018-08-23 14:41 GMT+02:00 Sergei Kornilov <sk@zsrv.org>: > >> Hello >> >> > This routine works well on postgres v10: >> it works? Or is only accepted as input? I got error on function call. >> I thought plpgsql cannot directly uses savepoints, only implicitly with >> begin ... exception ... end block ( https://www.postgresql.org/doc >> s/current/static/plpgsql-control-structures.html#PLPGSQL-ERROR-TRAPPING ) >> >> Well, you can still create this function with set check_function_bodies >> to off >> > > This doesn't help, because any function is validated before execution > but it is hard to believe, so it was working ever. > Regards > > Pavel > >> >> regards, Sergei >> >> >
-
RE: 11 beta 3 / ROLLBACK TO SAVEPOINT regression in PLPGSQL
David Klika <david.klika@atlas.cz> — 2018-08-23T14:45:02Z
> This routine works well on postgres v10: it works? Or is only accepted as input? I got error on function call. I thought plpgsql cannot directly uses savepoints, only implicitly with begin ... exception ... end block ( https://www.postgresql.org/docs/current/static/plpgsql-control-structures.html#PLPGSQL-ERROR-TRAPPING ) Well, you can still create this function with set check_function_bodies to off This doesn't help, because any function is validated before execution but it is hard to believe, so it was working ever. You are right, calling the method lead to error in both cases. My misunderstanding, sorry. David
-
Re: 11 beta 3 / ROLLBACK TO SAVEPOINT regression in PLPGSQL
Pavel Stehule <pavel.stehule@gmail.com> — 2018-08-23T14:57:48Z
2018-08-23 16:45 GMT+02:00 David Klika <david.klika@atlas.cz>: > > > This routine works well on postgres v10: > it works? Or is only accepted as input? I got error on function call. > I thought plpgsql cannot directly uses savepoints, only implicitly with > begin ... exception ... end block ( https://www.postgresql.org/ > docs/current/static/plpgsql-control-structures.html#PLPGSQL-ERROR-TRAPPING > ) > > Well, you can still create this function with set check_function_bodies to > off > > > > This doesn't help, because any function is validated before execution > > > > but it is hard to believe, so it was working ever. > > > > > > You are right, calling the method lead to error in both cases. My > misunderstanding, sorry. David > No problem :) Pavel