Re: [HACKERS] Definitional questions for pg_sequences view

2017-07-24 Thread Peter Eisentraut
On 7/20/17 16:36, Tom Lane wrote: > What exactly is the point of the new pg_sequences view? It is analogous to pg_tables, pg_matviews, pg_indexes, and other such system views that are sort of half-way between system catalogs and information schema. -- Peter Eisentraut

Re: [HACKERS] Definitional questions for pg_sequences view

2017-07-20 Thread Craig Ringer
On 20 July 2017 at 22:36, Tom Lane wrote: > > This could be fixed if it were possible to translate to > select * from pg_sequences where seqoid = 'my_seq'::regclass; > but the view isn't exposing the sequence OID. Should it? > It probably should. It's not part of

[HACKERS] Definitional questions for pg_sequences view

2017-07-20 Thread Tom Lane
What exactly is the point of the new pg_sequences view? It seems like it's intended to ease conversion of applications that formerly did "select * from sequencename", but if so, there are some fairly annoying discrepancies. The old way got you these columns: regression=# \d s1