Question

As the title suggests, my short question is : When using a dynamically loaded C function with a postgres trigger, is that a blocking or non-blocking call?

Background:
I need to generate some XML following an insert/update/delete on my database. The two options I’ve considered are:

1. Create some event table which is written to using during an I/U/D operation and have a daemon poll for these events and generate XML
2. Create a C function to generate this XML as part of a trigger in postgres.

Using option 2 would provide real-time updates but my concern is that there will be some overhead in generating the XML message and transmitting it etc, so inserting 100rows would take a lot longer due to creating an xml between each operation.

Answer 1

Yes, it “blocks”. The SQL that causes the trigger to run does not complete until the trigger is done.

See more info in the docs.