PostgreSQL 9.5 Released | Hacker News Reader

PostgreSQL 9.5 Released

postgresql.org

186 pointsgdeglin19 comments

You've jumped the gun. Postgres 9.5 is not due to be released until January the 7th. This link is to a stamp commit for 9.5.0; a git tag REL9_5_0 will be created separately.

Granted, it is true that it's been determined exactly what will be in the release. But packages have yet to be cut, and there is a media kit that has yet to be released. We'd have preferred if you'd waited until the 7th, which is slated as the official release date.

It is worth noting that while extremely rare, it is possible that we could encounter an issue during packaging which might necessitate a change, in which case that commit stamp might not be the same. Unless folks are comfortable working directly from git, it is probably best to wait for the official release.

> We'd have preferred if you'd waited until the 7th

Ok, let's bury this thread so that we can have a proper thread on the 7th. There's no hardship in waiting a couple days.

Thanks!

Hope I was not unduly curt with the poster, but there are simply no packages available for 9.5.0 today, and won't be until Thursday. There is an official embargo observed by the -packagers group.

If you're curious about what's new in Postgres 9.5 give https://wiki.postgresql.org/wiki/What's_new_in_PostgreSQL_9.... or http://www.craigkerstiens.com/2015/12/27/postgres-9-5-featur... a look.

Postgres 9.5 is a great release. Upsert is, obviously, a very welcome addition.

But I am more interested in row level security policies and its impact on evolution of technologies like Postgrest. It may even spawn a new generation of backend architectures that don't really need app servers to power SPAs and mobile apps. A possible exciting new niche for Postgres.

What is not great about Postgres ecosystem is that megacorps who benefit the most from selling hosted Postgres instances ( AWS/Amazon and Compose.io/IBM ) are not present in the list of significant sponsors of PostgreSQL foundation! Heroku/Salesforce is only on the Silver sponsorship level.

That's really bad.

http://www.postgresql.org/about/sponsors/

Probably true of most open source projects, whether they have a foundation or not.

Took the extent that amazon sponsors open source, I suspect that'll change as AWS RDS grows in importance...

Amazon RDS support for PG was released in November 2013 so I wouldn't hold my breath.

I'm really glad to know that this release is coming out in the next few days. I'm going to start building a new multitenant application soon and I'd love to be able to try row-level security.

It seems that the sponsorship levels are determined by years of sponsorship [1]. Not really defending these specific companies as they have been profiting from PostgreSQL for a lot longer than one year. However, they could be on track to become Platinum sponsors and we wouldn't know.

[1] https://wiki.postgresql.org/wiki/Sponsoring

It looks like PostgREST has already started updating its documentation for RLS: https://github.com/begriffs/postgrest

Release notes: http://www.postgresql.org/docs/devel/static/release-9-5.html

This is especially great news for me personally.

I'm developing an app in PHP 7 that hasn't been released yet, and I was hoping to make it use PostgreSQL 9.5 features (i.e. modifying JSONB). But I was worried about it not being released before my project was finished.

Luckily, I was significantly delayed!

EDIT: Looks like it actually comes out on the 7th, but my delays are pretty significant.

Anyone have any idea when we're likely to see this on RDS (or Google SQL, for that matter)?

if past precedent is any indicator, about 6 months from now for RDS

I thought Google SQL was based on MySQL?

It is; this is a not-so-subtle feature request on my part to Google.

Does postgres have any use at Google? When I was there, they had many ex-mysql people and used mysql internally (e.g. for YouTube, iirc.)

pg_rewind is great, with this the infrastructure for some really great replication tools are in place - recovering a failed primary has been my biggest pain point with my repmgr + barman setup, so combined with some UI improvements (reducing the incantations needed to initiate a failover and restore a failed primary) I'd say it could be about as painless as setting up AlwaysOn Availability Groups with MS SQL Server.

Reply on news.ycombinator.com