[nonrel] Factor out has_result into database backend

[Jonas H.]

(This patch is part of the changes made to the inofficial fork "Django-nonrel" which adds basic support for non-relational databases.)

Refactors some code of sql.query.Query.has_result into the database backend. This is required so that database backends can use specialized more performant "query has any results" checks.

[Carl Meyer]

Giving backends more control here makes sense to me.

[Jonas H.]

Does this need tests?

[Carl Meyer]

Pretty sure the test suite already exercises this code quite heavily, so I think in this case of a pure refactoring there aren't more tests needed, as long as the existing ones pass. And obviously docs aren't relevant at this level either.

In terms of the patch specifics, what's the justification for why some query modification bits are moved into the compiler's has_results and not others?

[Jonas H.]

I have no idea. Maybe I can get the original author, Waldemar Kornewald, to join these threads...

[Waldemar Kornewald]

The reason why I only moved two lines to the compiler is that all the other operations are DB independent optimizations. The selected fields and query ordering etc need to be cleared on NoSQL, too. Only the extra mask is a SQL specific optimization and other DBs might use some other trick. Anyway,it would be easy to change the patch to also allow for overriding the other operations, but we should keep the current separation between extra mask and general optimizations because all nonrel backends only override the extra mask code and they shouldn't have to reimplement the general optimizations unnecessarily.

[Tim Graham]

Duplicate of #14857 which was closed as "won't fix" by Russ, but I'm not sure if his reasoning still holds since query-refactor was never merged and NoSQL support seems to have dropped off the radar as far as I know:

"NoSQL support isn't something that needs to be addressed piecemeal. This isn't the only place that raw SQL leaks out of the compiler; these leaks need to be addressed as a whole, not piecemeal. As I've told you *many* times Waldemar, NoSQL support is on my radar for 1.4, when we address merging Alex's query-refactor SoC branch."

I've updated the patch to apply cleanly and ensure that all tests pass. Any objections to merging this?

​https://github.com/django/django/pull/1346

[Russell Keith-Magee]

The NoSQL branch efforts are obviously dead, but the pull request here makes perfect sense to me. It's a small piece of the puzzle, but it doesn't hurt, so no objections on my part to merging.

[Tim Graham <timograham@…>]

In 95eb68c98fe2dc95277d4e26a0316d856dc868dd:

Fixed #17339 -- Factor out has_result into database backend.

Thanks jonash.