Обсуждение: BUG #9749: ERROR: unexpected classid 3600
The following bug has been logged on the website: Bug reference: 9749 Logged by: clime Email address: clime7@gmail.com PostgreSQL version: 9.2.4 Operating system: linux Description: I am getting this error when trying to execute "reassign owned" command. cb_test=# reassign owned by clime to cb_test; ERROR: unexpected classid 3600 cb_test=# select '3600'::regclass; regclass ------------ pg_ts_dict (1 row)
clime7@gmail.com wrote: > I am getting this error when trying to execute "reassign owned" command. > > cb_test=# reassign owned by clime to cb_test; > ERROR: unexpected classid 3600 > cb_test=# select '3600'::regclass; > regclass > ------------ > pg_ts_dict Hmm, The code is clearly missing handling that case, as well as all text search objects types. Will fix, thanks for the report! -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
Alvaro Herrera wrote: > clime7@gmail.com wrote: > > > I am getting this error when trying to execute "reassign owned" command. > > > > cb_test=# reassign owned by clime to cb_test; > > ERROR: unexpected classid 3600 > > cb_test=# select '3600'::regclass; > > regclass > > ------------ > > pg_ts_dict > > Hmm, The code is clearly missing handling that case, as well as all text > search objects types. Will fix, thanks for the report! Oops, I just noticed that the fix is very simple in HEAD and 9.3 because we can just add the appropriate case labels to the already existing generic object alter-owner case. But in 9.2, we not only need to add extra cases to cover each object, but we also need to refactor the appropriate AlterOwner routine to work on OID input rather than name only. In other words we need the equivalent of 0c7b9dc7d037c4465227dc2300ff48019feeba37 for each of the text search object types :-( Not sure I have the time to do all that right now. Contributions welcome. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> Oops, I just noticed that the fix is very simple in HEAD and 9.3 because
> we can just add the appropriate case labels to the already existing
> generic object alter-owner case. But in 9.2, we not only need to add
> extra cases to cover each object, but we also need to refactor
> the appropriate AlterOwner routine to work on OID input rather than name
> only. In other words we need the equivalent of
> 0c7b9dc7d037c4465227dc2300ff48019feeba37 for each of the text search
> object types :-(
> Not sure I have the time to do all that right now. Contributions
> welcome.
Given the lack of complaints to date, maybe fixing it in 9.3/HEAD is
sufficient. I'd certainly rather see those branches get fixed now,
and 9.2 later, than nothing happen at all for a long time. It's also
arguable that back-patching such a large change into 9.2 is more risk
than the problem is worth.
regards, tom lane
Tom Lane wrote: > Alvaro Herrera <alvherre@2ndquadrant.com> writes: > > Oops, I just noticed that the fix is very simple in HEAD and 9.3 because > > we can just add the appropriate case labels to the already existing > > generic object alter-owner case. But in 9.2, we not only need to add > > extra cases to cover each object, but we also need to refactor > > the appropriate AlterOwner routine to work on OID input rather than name > > only. In other words we need the equivalent of > > 0c7b9dc7d037c4465227dc2300ff48019feeba37 for each of the text search > > object types :-( > > > Not sure I have the time to do all that right now. Contributions > > welcome. > > Given the lack of complaints to date, maybe fixing it in 9.3/HEAD is > sufficient. I'd certainly rather see those branches get fixed now, > and 9.2 later, than nothing happen at all for a long time. It's also > arguable that back-patching such a large change into 9.2 is more risk > than the problem is worth. Patching 9.3 and up only was my first thought, but I then noticed that the reporter is using 9.2. I'm not opposed to doing 9.3 right now and older branches later, except that I'm afraid it'd fall by the wayside and we'd never do it. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
Tom Lane wrote: > Alvaro Herrera <alvherre@2ndquadrant.com> writes: > > Oops, I just noticed that the fix is very simple in HEAD and 9.3 because > > we can just add the appropriate case labels to the already existing > > generic object alter-owner case. But in 9.2, we not only need to add > > extra cases to cover each object, but we also need to refactor > > the appropriate AlterOwner routine to work on OID input rather than name > > only. In other words we need the equivalent of > > 0c7b9dc7d037c4465227dc2300ff48019feeba37 for each of the text search > > object types :-( > > > Not sure I have the time to do all that right now. Contributions > > welcome. > > Given the lack of complaints to date, maybe fixing it in 9.3/HEAD is > sufficient. I'd certainly rather see those branches get fixed now, > and 9.2 later, than nothing happen at all for a long time. It's also > arguable that back-patching such a large change into 9.2 is more risk > than the problem is worth. Just pushed a fix for all branches. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services