But to answer your question the reason we need autocommit off is because PostgreSQL would have to use a WITH HOLD cursor outside of a transaction. WITH HOLD cursors materialize cursors to disk. There is no technical reason we don't support them. If you are interested you can supply a PR for us.
If you have any question related to jdbc driver you should send it to pgsql-jdbc@postgresql.org . Also I would like to mention that you could use defaultRowFetchSize = int at connection level or setFetchSize at statement level to define how many record will be read by resultset .
Bug reference: 15152 Logged by: Isaac Arasavalli Email address: aistrong.hs@gmail.com PostgreSQL version: 10.1 Operating system: Windows 7 Description:
When I fetch a table with huge data, the result set fails with heap issue with postgreSql. But in DB2 it fetches group by group so it handles. To fix there is an option to set autocommit off and set fetch size at resultset.
For, this I have to do lot of code change due to many occurrences of simila issue in my huge application.
Is there any alternative to fix this, like setting any parameters at data source leve or any other way to minimize code change and solve.
