Re: proposal: make NOTIFY list de-duplication optional

Filip Rembiałkowski <filip.rembialkowski@gmail.com>

From: Filip Rembiałkowski <filip.rembialkowski@gmail.com>
To: Merlin Moncure <mmoncure@gmail.com>
Cc: Pgsql Hackers <pgsql-hackers@postgresql.org>
Date: 2016-02-09T19:16:53Z
Lists: pgsql-hackers
On Tue, Feb 9, 2016 at 12:15 AM, Merlin Moncure <mmoncure@gmail.com> wrote:

> I wonder if the third argument
> should be a boolean however.  If we make it 'text, 'send mode',
> instead, we could leave some room for more specialization of the
> queuing behavior.
>
> For example, we've had a couple of requests over the years to have an
> 'immediate' mode which dumps the notification immediately to the
> client without waiting for tx commit. This may or may not be a good
> idea, but if it was ultimately proved to be, it could be introduced as
> an alternate mode without adding an extra function.

But then it becomes disputable if SQL syntax change makes sense.

---we had this,
 NOTIFY channel [ , payload ]
---and in this patch we have this
NOTIFY [ ALL | DISTINCT ] channel [ , payload ]
 ---  but maybe we should have this?
NOTIFY channel [ , payload [ , mode ] ]

I'm not sure which direction is better with non-standard SQL additions.
Recycling keywords or adding more commas?


Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →
  1. Use a hash table to de-duplicate NOTIFY events faster.