Fix propagating post-save mutations

[booleangate]

This addresses the bug I reported in #2523. I've broken commits up so you can review just the first two in isolation (a failing test, and the fix) to cut out all the noise from the regeneration.

📓 I had to branch my changes off of the fix/bump-atlas branch in order to be able to run go generate ./.... I'm just waiting for that to merge, then I'll rebase.

[a8m]

Thanks for the contribution @booleangate! The changes look good.

Please, fix the minor suggestions and I'll merge this PR. Also note, that the changes need to be applied on the other hook types as well (update, update-one, etc). Would you like to contribute them in different PRs?

[a8m]

💯

[booleangate]

Also note, that the changes need to be applied on the other hook types as well (update, update-one, etc). Would you like to contribute them in different PRs?

@a8m I'm afraid I don't follow but I'm happy to learn. The signature on the update and delete clients is Save(...) (int, error) and Exec(...) (int, error) respectively. Because they return affected row count and not the model itself, I don't see how we could make a similar change to those clients.

[a8m]

Also note, that the changes need to be applied on the other hook types as well (update, update-one, etc). Would you like to contribute them in different PRs?

@a8m I'm afraid I don't follow but I'm happy to learn. The signature on the update and delete clients is Save(...) (int, error) and Exec(...) (int, error) respectively. Because they return affected row count and not the model itself, I don't see how we could make a similar change to those clients.

Let's start with update-one - https://github.com/ent/ent/blob/master/entc/gen/template/builder/update.tmpl#L200-L202

[a8m]

CI is red. Please, run go generate ./... from the root of the project. Thanks 🙏

[a8m]

@booleangate can you rebase your branch on master and create a new PR?

[booleangate]

@a8m yup, working on a new PR now with the UpdateOne changes.