remove support for Prefer: params=single-object

[joelonsql]

Using this preference was deprecated in 6c3d7a9, in favor of Functions with an array of JSON objects.

This change removes support for this feature, planned for the next major version.

[wolfgangwalther]

Eyeballing it, it LGTM. CI is also green. Should be good to go.

[wolfgangwalther]

I think this is similar to the removal of PG 9.6 / 10 / 11 support and could go to "Changed".

The commits for that were done with feat: prefix, so I guess this one should be as well. Yeah.. it's quite the opposite of a "new feature", but it's not any of the prefixes we use either ;)

[steve-chavez]

Nice work 👍

[wolfgangwalther]

I don't think the CHANGELOG was addressed properly. Now we have a Removed section at the top. We didn't have any removed sections before and if so we should have it at the bottom. In any case, the PG removal should be in the same section. As I wrote above, this should just be in the "Changed" section.

Also, @steve-chavez, I think we loosely agreed on some form of conventional commits, i.e. using prefixes. The squashed commit message has none.

[steve-chavez]

Ah, my bad, I thought it was addressed. How about this?

• I revert the previous commit.
• Edit the CHANGELOG.
• Make a commit with the break: .. prefix. We had some of these in the past.

All in one push to main?

[steve-chavez]

Also, @steve-chavez, I think we loosely agreed on some form of conventional commits, i.e. using prefixes. The squashed commit message has none.

Btw, maybe we can enforce this somehow? https://dev.to/mbarzeev/a-git-hook-for-commit-messages-validation-no-husky-just-js-1hni

I guess squash would have to be disabled too.

[wolfgangwalther]

I don't think you need to revert. I'd say it's fine to just do the changelog in a follow up commit.

We'd have to think about the proper commit prefix for something like this. I used feat: for the old PG versions removal, but maybe we should just introduce remove:?

Btw, maybe we can enforce this somehow?

Yeah, I use commitlint in a different project. Works well. We'd have to standardize on the prefixes first, though ;)

[wolfgangwalther]

I guess squash would have to be disabled too.

No, I think squash is fine. You can still edit the commit message when squashing.

[steve-chavez]

We'd have to think about the proper commit prefix for something like this. I used feat: for the old PG versions removal, but maybe we should just introduce remove:?

remove: LGTM.

Fixed the changelog on b821857