Обсуждение: How to get RelationName ??
Hello, How can i get the name of a relation from its Oid ?? I have seen some functions in utils/cache/relcache.h like RelationIdGetRelation - which gives Relation Node from Oid RelationSysNameGetRelation - which gives Relation node for only the System names. Can anyone please tell me how to get Relation node from relation name ??? Thanks, -Ramu _________________________________________________________________ Contact brides & grooms FREE! http://www.shaadi.com/ptnr.php?ptnr=hmltag Only on www.shaadi.com. Register now!
On Wednesday 10 March 2004 09:58 am, Ramanujam H S Iyengar wrote: > Hello, > > How can i get the name of a relation from its Oid ?? I have seen some > functions in utils/cache/relcache.h like RelationIdGetRelation - which > gives Relation Node from Oid RelationSysNameGetRelation - which gives > Relation node for only the System names. > > Can anyone please tell me how to get Relation node from relation name ??? > Won't you have to go get the data from the database for this? I mean actually execute the query: select relname from pg_class where oid = 1234; -- Jonathan Gardner jgardner@jonathangardner.net
Hello, Sorry for the previous confusing mail .. iam in need of a function through which i can get a Relation Node given its name .. precisely the same one equivalent to " RelationSysNameGetRelation " .. which works for all types of relations (system and user tables) On Wednesday 10 March 2004 09:58 am, Ramanujam H S Iyengar wrote: >Hello, > >How can i get the name of a relation from its Oid ?? I have seen some >functions in utils/cache/relcache.h like RelationIdGetRelation - which >gives Relation Node from Oid RelationSysNameGetRelation - which gives >Relation node for only the System names. > >Can anyone please tell me how to get Relation node from relation name ??? > Won't you have to go get the data from the database for this? I mean actually execute the query: Iam trying to test for some patch in the code of the Optimizer(planner) and so I need some C functions, which does the job. I think we must have some functions like the ones i have mentioned for the purpose. isin't it ?? regards, -Ramu >select relname from pg_class where oid = 1234; >-- >Jonathan Gardner >jgardner ( at ) jonathangardner ( dot ) net _________________________________________________________________ Need a job? Get head-hunted by the best. http://www.naukri.com/msn/index.php?source=hotmail Post your CV free!
"Ramanujam H S Iyengar" <hals_ramu@hotmail.com> writes:
> iam in need of a function through which i can get a Relation Node given its
> name ..
> precisely the same one equivalent to " RelationSysNameGetRelation " .. which
> works for all types of relations (system and user tables)
The reason RelationSysNameGetRelation only works for system relations is
that it assumes the schema name is "pg_catalog". For non-system
relations you cannot do a lookup based only on relation name, because
there's no certainty that the result is unique.
There are some functions in catalog/namespace.c that can do lookup of an
unqualified name relative to the current schema search path, if that's
what you want.
> Iam trying to test for some patch in the code of the Optimizer(planner) and
> so I need some C functions, which does the job.
The optimizer has no need, ever, to find a relation by name; all it ever
sees are predigested relation OIDs. So you are not making a lot of
sense here. You certainly cannot assume that a search-path lookup is
appropriate for a relation that the optimizer is dealing with.
regards, tom lane
>>"Ramanujam H S Iyengar" <hals_ramu ( at ) hotmail ( dot ) com> writes: >>iam in need of a function through which i can get a Relation Node given >>its name .. >>precisely the same one equivalent to " RelationSysNameGetRelation " .. >>which works for all types of relations (system and user tables) >The reason RelationSysNameGetRelation only works for system relations is >that it assumes the schema name is "pg_catalog". For non-system >relations you cannot do a lookup based only on relation name, because >there's no certainty that the result is unique. >There are some functions in catalog/namespace.c that can do lookup of an >unqualified name relative to the current schema search path, if that's >what you want. Thanks a lot, The function RelnameGetRelid(const char *) is working for me !! >>Iam trying to test for some patch in the code of the Optimizer(planner) >>and so I need some C functions, which does the job. >The optimizer has no need, ever, to find a relation by name; all it ever >sees are predigested relation OIDs. So you are not making a lot of >sense here. You certainly cannot assume that a search-path lookup is >appropriate for a relation that the optimizer is dealing with. regards, tom lane Iam trying to put in some other module of optimizer that we have developed, which works on relation name and gives the optimal plan. Iam trying to convert the plans given by this(our) optimizer to the Plan structure of PostgreSQL. So finally i have to convert the Relation names back to their relOids. All this in the process of using the execution engine of Postgres for our optimizer module !! :-) Thats why i was asking !! Thanks -Ramu _________________________________________________________________ Take a loan. Win great prizes! Handsome prizes to be won! Take a loan & win TV, Fridge & many more prizes ! http://go.msnserver.com/IN/44044.asp
"Ramanujam H S Iyengar" <hals_ramu@hotmail.com> writes:
>> The optimizer has no need, ever, to find a relation by name; all it ever
>> sees are predigested relation OIDs. So you are not making a lot of
>> sense here. You certainly cannot assume that a search-path lookup is
>> appropriate for a relation that the optimizer is dealing with.
> Iam trying to put in some other module of optimizer that we have developed,
> which works on relation name and gives the optimal plan. Iam trying to
> convert the plans given by this(our) optimizer to the
> Plan structure of PostgreSQL. So finally i have to convert the Relation
> names back to their relOids.
If your optimizer emits unqualified relation names then it is broken,
as it will never be safe to use in the presence of schemas. People
will not want to use a database that might apply updates meant for
"a.foo" to "b.foo".
regards, tom lane