It will be hard to supply a reproducer since it seems to require lots of concurrent startups, but I will see if I can
simplifythe reproducible case.
Your explanation seems plausible. I will add the suggested logging and rebuild and test to log the oid and other
information.It will probably be a few days before I have those results.
Thanks for your suggestions and time,
Zach Marshall
-----Original Message-----
From: Vladimir Sitnikov [mailto:sitnikov.vladimir@gmail.com]
Sent: Wednesday, February 17, 2016 9:20 AM
To: Zachary Marshall <Zachary.Marshall@sas.com>
Cc: pgsql-jdbc@postgresql.org
Subject: Re: [JDBC] NullPointerException in TypeInfoCache.getSQLType
Do you have a reproducer?
Can you check if you have types with typname of null?
select * from pg_catalog.pg_type t where t.typname is null
> all are typically creating new schemas
Can it be like:
1) org.postgresql.jdbc.PgDatabaseMetaData#getTypeInfo loads all the type ids
2) as it reaches org.postgresql.jdbc.TypeInfoCache#getPGType(int), the type is no longer in the DB (e.g. dropped) ?
Can you add logging to TypeInfoCache#getPGType & rebuild & test to log the actual oid in question?
Vladimir
