Implement message headers

[v0idpwn]

Closes #203
Needed for #294

[theory]

Looks straightforward, but please:

• Add an upgrade script that can upgrade the SQL from the previous version to this version's change. Details here.
• Add tests for the new functions and behaviors.

Thanks!

[v0idpwn]

Of course, this is an API proposal, not ready yet :)
Sorry, communicated with @ChuckHend but didn't write it in description

[theory]

Of course, this is an API proposal, not ready yet :) Sorry, communicated with @ChuckHend but didn't write it in description

Ah, got it, didn't realize, thanks for clarifying.

[theory]

(Consider converting this to a Draft PR)

[theory]

Consider replacing the variants with just two using defaults:

-- send: 4 args with integer delay
CREATE FUNCTION pgmq.send(
    queue_name TEXT,
    msg JSONB,
    headers JSONB DEFAULT = NULL,
    delay INTEGER DEFAULT = 0
) RETURNS SETOF BIGINT AS $$
    SELECT * FROM pgmq.send(queue_name, msg, headers, clock_timestamp() + make_interval(secs => delay));
$$ LANGUAGE sql;

CREATE FUNCTION pgmq.send(
    queue_name TEXT,
    msg JSONB,
    headers JSONB DEFAULT = NULL,
    delay TIMESTAMP WITH TIME ZONE = clock_timestamp(),
) RETURNS SETOF BIGINT AS $$
    SELECT * FROM pgmq.send(queue_name, msg, headers, delay);
$$ LANGUAGE sql;

[v0idpwn]

I think it's not equivalent. I'm accepting pgmq.send(queue_name, msg, integer/timestamp) too

[theory]

Again, try using DEFAULT expressions in parameter lists to reduce the number of variants.

[academiaresf]

i can help for merging this?

[ChuckHend]

i can help for merging this?

We do need some tests!