Re: pg_standby for 8.2 (with last restart point) - Mailing list pgsql-hackers

On Fri, 28 Mar 2008, Gurjeet Singh wrote:

>> This project doesn't make functional changes to stable releases, that's
>> the reason why 8.2 will never get patched to add the %r feature.
> I completely understand that, but still was hoping that we'd change that.

Well, then you really don't understand this at all then, so let's work on
that for a bit.  http://www.postgresql.org/support/versioning is the
official statement, perhaps some examples will help clarify where and why
the line is where it is.

One of the first patches I ever submitted made a minor change to a contrib
utility used solely for benchmarking (pgbench) that added a useful
feature, only if you passed it the right parameter.  That was considered
for a tiny bit before being rejected as a feature change too large to put
into a stable branch.

That was a small change in a utility that should never be run on a
production system.  You're trying to get a change made to the code path
people rely on for their *backups*.  Good luck with that.

The parable I enjoy pulling out in support of this policy is MySQL bug
#31001:

http://www.mysqlperformanceblog.com/2007/10/04/mysql-quality-of-old-and-new-features/

where they added a seemingly minor optimization to something and
accidentally broke the ability to sort in some cases.  There's always a
small risk that comes with any code change, and this is why you don't ever
touch working production code unless you're fixing a bug that's more
troublesome than that risk.