I don't think that's an accurate characterization. I'm sure there are some improvements to hinting that would be approved if they are well thought out and don't introduce major maintainability concerns, etc. Someone just needs to do the work required, including considering many alternative approaches which may be radically different from what you typically think of as hints.
The proposals that tend to fail are ones that follow an argument pattern like "Oracle does it this way... therefore it is good... therefore postgres should do something similar".
I'm not trying to dismiss your point at all. I think it's important that postgres give users as much power as possible, and a part of that is having some control over the optimizer without waiting for a new release. Postgres offers that in some innovative ways, but it could do better.
Index hints already are supported in multiple different ways:
http://thoughts.davisjeff.com/2011/02/05/why-postgresql-alre...
"I see this as a case of devs overriding users"
I don't think that's an accurate characterization. I'm sure there are some improvements to hinting that would be approved if they are well thought out and don't introduce major maintainability concerns, etc. Someone just needs to do the work required, including considering many alternative approaches which may be radically different from what you typically think of as hints.
The proposals that tend to fail are ones that follow an argument pattern like "Oracle does it this way... therefore it is good... therefore postgres should do something similar".
I'm not trying to dismiss your point at all. I think it's important that postgres give users as much power as possible, and a part of that is having some control over the optimizer without waiting for a new release. Postgres offers that in some innovative ways, but it could do better.