Gavin Sherry wrote:
>I was initially against the idea of using libpq but its growing on me too.
>
>I think it would be good if the core functions of pg_autovacuum: threshold
>algorithms, connection, issuing commands can be (re?)designed such that
>not only the backend can link against it but also a stripped down
>pg_autovacuum binary which can be used manually. That way users can have a
>choice and a workaround if there are any issues with the backend model.
>Also useful for debugging.
>
I agree. Initially, it appears that this won't be a problem since the
pg_autovacuum executable will be untouched (or as untouched as possible
anyway), it will just be launched by the backend. Going forward if I
use any of the functionality provided by the backend (error reporting
etc...) I will then have to deign it to work in both stand alone and
postmaster sub-process modes, which I think is doable.
Matthew