Peter Eisentraut <peter_e@gmx.net> writes:
> On 10/9/14 1:58 PM, Fujii Masao wrote:
>> Also I think that it's useful to allow ALTER ROLE/DATABASE SET to
>> set PGC_BACKEND and PGC_SU_BACKEND parameters. So, what
>> about applying the attached patch? This patch allows that and
>> changes the context of session_preload_libraries to PGC_SU_BACKEND.
> After looking through this again, I wonder whether there is any reason
> why ignore_system_indexes cannot be plain PGC_USERSET. With this
> change, we'd allow setting it via ALTER ROLE, but the access to
> pg_db_role_setting happens before it. So if there is anything unsafe
> about changing ignore_system_indexes, then this would be a problem, but
> I don't see anything.
There are some, um, "interesting" consequences of setting
ignore_system_indexes; AFAIK, none that put data integrity at risk, but it
can destroy performance in ways beyond the obvious ones. See for example
the comments for get_mergejoin_opfamilies and get_ordering_op_properties.
I don't particularly want to answer user bug reports about such behaviors,
nor do I care to put any effort into making the behavior without system
indexes smarter than it is now. (We should also consider the risk that
there might be as-yet-unrecognized dependencies on catalog scan order that
would amount to actual bugs.)
So I'm -1 on any change that might make it look like we were encouraging
people to use ignore_system_indexes except as a very last resort.
regards, tom lane
