Thread: missing event trigger support functions in 9.3
Hello
I am writing a article about 9.3. I found so event trigger functions is not complete. We have only pg_event_trigger_dropped_objects() function. It looks really strange and asymmetric. Can we implement similar function for CREATE as minimum to 9.3?
I am expecting so this function should not be too complex - and can be moved to contrib maybe (if it is too late now).I am writing a article about 9.3. I found so event trigger functions is not complete. We have only pg_event_trigger_dropped_objects() function. It looks really strange and asymmetric. Can we implement similar function for CREATE as minimum to 9.3?
Pavel
On 2013.05.09 10:40 AM, Pavel Stehule wrote: > I am writing a article about 9.3. I found so event trigger functions is not > complete. We have only pg_event_trigger_dropped_objects() function. It looks > really strange and asymmetric. Can we implement similar function for CREATE as > minimum to 9.3? > > I am expecting so this function should not be too complex - and can be moved to > contrib maybe (if it is too late now). Really, the touted new event triggers feature in fact only works for dropping objects rather than for all DDL or other events? If so, then the feature seems seriously incomplete. Are drops the main expected use case for the feature, rather than "tell me when something happened"? -- Darren Duncan
Darren Duncan <darren@darrenduncan.net> writes: > On 2013.05.09 10:40 AM, Pavel Stehule wrote: >> I am expecting so this function should not be too complex - and can be moved to >> contrib maybe (if it is too late now). My understanding is that it's too late even for contrib now. > Really, the touted new event triggers feature in fact only works for > dropping objects rather than for all DDL or other events? If so, then the > feature seems seriously incomplete. Are drops the main expected use case > for the feature, rather than "tell me when something happened"? The feature is seriously incomplete in 9.3, though not in the way you describe here. To simplify things, please consider that all we have now is code infrastructure so that we can make progress in the next releases. Maybe. Regards, -- Dimitri Fontaine http://2ndQuadrant.fr PostgreSQL : Expertise, Formation et Support
2013/5/9 Dimitri Fontaine <dimitri@2ndquadrant.fr>
>> I am expecting so this function should not be too complex - and can be moved toMy understanding is that it's too late even for contrib now.
>> contrib maybe (if it is too late now).The feature is seriously incomplete in 9.3, though not in the way you
> Really, the touted new event triggers feature in fact only works for
> dropping objects rather than for all DDL or other events? If so, then the
> feature seems seriously incomplete. Are drops the main expected use case
> for the feature, rather than "tell me when something happened"?
describe here. To simplify things, please consider that all we have now
is code infrastructure so that we can make progress in the next
releases. Maybe.
I understand and I know, so current state is a maximum what is possible, but it looks really strange
Pavel
Regards,
--
Dimitri Fontaine
http://2ndQuadrant.fr PostgreSQL : Expertise, Formation et Support
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Darren Duncan wrote: > On 2013.05.09 10:40 AM, Pavel Stehule wrote: > >I am writing a article about 9.3. I found so event trigger functions is not > >complete. We have only pg_event_trigger_dropped_objects() function. It looks > >really strange and asymmetric. Can we implement similar function for CREATE as > >minimum to 9.3? > > > >I am expecting so this function should not be too complex - and can be moved to > >contrib maybe (if it is too late now). > > Really, the touted new event triggers feature in fact only works for > dropping objects rather than for all DDL or other events? It works for all DDL. It provides complete information for drops, not other kinds of commands. This will be expanded later. > If so, then the feature seems seriously incomplete. Feel free to contribute during the 9.4 development cycle. Your appreciation for the work that already went into this feature is duly noted. > Are drops the main expected use case for the feature, rather than > "tell me when something happened"? No. It was what we had time to implement. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
2013/5/9 Pavel Stehule <pavel.stehule@gmail.com>
RegardsHelloI am expecting so this function should not be too complex - and can be moved to contrib maybe (if it is too late now).
I am writing a article about 9.3. I found so event trigger functions is not complete. We have only pg_event_trigger_dropped_objects() function. It looks really strange and asymmetric. Can we implement similar function for CREATE as minimum to 9.3?
Pavel
Maybe as workaround can help:
CREATE OR REPLACE FUNCTION snitch() RETURNS event_trigger AS $$
DECLARE
my_cmd text;
BEGIN
SELECT query FROM pg_stat_activity where pid = pg_backend_pid() INTO my_cmd;
RAISE NOTICE 'snitch: % % %', tg_event, tg_tag, my_cmd;
END;
$$ LANGUAGE plpgsql;
CREATE EVENT TRIGGER snitch ON ddl_command_start EXECUTE PROCEDURE snitch();
Then based on tg_tag - you can have different rules to take object_name from my_cmd - or whatever else is needed for additional rules...
Kind Regards,
Misa
On 2013.05.09 11:22 AM, Alvaro Herrera wrote: > Darren Duncan wrote: >> On 2013.05.09 10:40 AM, Pavel Stehule wrote: >>> I am writing a article about 9.3. I found so event trigger functions is not >>> complete. We have only pg_event_trigger_dropped_objects() function. It looks >>> really strange and asymmetric. Can we implement similar function for CREATE as >>> minimum to 9.3? >>> >>> I am expecting so this function should not be too complex - and can be moved to >>> contrib maybe (if it is too late now). >> >> Really, the touted new event triggers feature in fact only works for >> dropping objects rather than for all DDL or other events? > > It works for all DDL. It provides complete information for drops, not > other kinds of commands. This will be expanded later. > >> If so, then the feature seems seriously incomplete. > > Feel free to contribute during the 9.4 development cycle. > > Your appreciation for the work that already went into this feature is > duly noted. Sorry if I came across as not appreciating the work already done. I in fact do appreciate it. I made a mistake. Please disregard my initial comment. -- Darren Duncan