conn: avoid try/catch blocks

Question

conn: avoid try/catch blocks

Closed this issue 8 years ago · 2 comments

Answer 1 · 2017-01-10T01:44:03.000Z

The PG documentation states that volatile-ness of data shared between try and catch blocks needs to be handled correctly. Currently we ignore this, but a better solution is to refactor some of the shared routines in connection.c to allow us to avoid the try/catch blocks.

Answer 2 · 2017-01-10T02:03:12.000Z

This was fixed by duplicating the get cluster connection logic and handling the errors explicitly. see connection.c for comments.