Re: Signature change for SPI_cursor_open - Mailing list pgsql-hackers

From Thomas Hallgren
Subject Re: Signature change for SPI_cursor_open
Date
Msg-id thhal-0BYJfAr/W3kAWrQhCjD4kA228g9TPwB@mailblocks.com
Whole thread Raw
In response to Re: Signature change for SPI_cursor_open  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
Tom Lane wrote:> You should set the flag if and only if you are executing a pl/java> function that has a provolatile
settingof "stable" or "immutable".> The new rule is that only functions declared "volatile" are allowed> to have side
effectson the database.  See pghackers discussions in> early September about updating snapshots, doing
CommandCounterIncrement,>and so forth within functions.>
 
Ok, now I understand. Thanks for the explanation. I guess that if 
read_only is set to true and an attempt is made to execute a plan that 
has side effect, that will cause an ERROR?

Regards,
Thomas Hallgren




pgsql-hackers by date:

Previous
From: Alvaro Herrera
Date:
Subject: Re: 8.0b4: COMMIT outside of a transaction echoes ROLLBACK
Next
From: Tom Lane
Date:
Subject: Re: Signature change for SPI_cursor_open