use pgx::prelude::*;
#699
use pgx::prelude::*;
#699
Conversation
Previously, PGX has relied on `use pgx::*;` in order to make the framework happen. However, this severely pollutes the namespaces of our users. We're hogging idents. Worse, a user collision with our identifiers may result in displacing our logic, resulting in undesirable and unexpected side effects. PGX depends on the macros being at least mostly correct in order to be sound, so this is not at all acceptable.
This macro is also fairly critical to PGX's functioning, being integral to any and all "table"-like returns.
`use pgx::prelude::*;` is already sufficient for some test files now,
|
Assuming PL/Rust works with |
|
@workingjubilee does that mean the user-code can no longer access all of pgx? Or just makes it easier/cleaner to detect |
|
Currently the latter, but it makes the former humanly possible without completely breaking... uh, everything. |
workingjubilee
added
the
refactor
|
I mentioned that a lot of PGX's logic is deeply entangled, and this PR involves sorting... most of that out. Before this PR, PGX has what amounts to an existential dependency on the glob-export that you do not enjoy. #615 made the situation significantly better, but it still remains a problem, even after this amount of refactoring. |
|
This now makes all of |
|
Well... except PL/Rust. |
As an alternative to exporting all identifiers from PGX, this PR introduces a module for importing all commonly-used, distinctive identifiers from PGX. It also makes it much less likely that a random name introduced by a user will shadow any of PGX's own names by fully qualifying the names used in various macro expansions.