Hasura 2.x docs state snake-cased pk_columns in mutations, yet Hasura 2.21 uses camelCase

[CuriousLearner]

Version Information

Server Version: 2.21

Environment

OSS

What is the current behaviour?

Upgrading Hasura from 2.16.0 to 2.21.0 broke the mutations, since the pk_columns in default mutations are referenced as camelCase pkColumns.

There is no mention in the release notes about this change.

Furthermore, a minor version upgrade should not have a breaking change.

What is the expected behaviour?

A minor version upgrade should not have breaking change in the API.

How to reproduce the issue?

1. Use mutation with pk_columns in version 2.16.0
2. Upgrade to 2.21.0
3. Now mutations uses pkColumns.

Any possible solutions/workarounds you're aware of?

Degrading back to 2.16.0

Keywords

breaking-change, camecase, snakecase, pk columns

[rikinsk]

Hey there were some fixes related to the naming conventions feature introduced in v2.21.

"Fix graphql-default naming convention bugs (#9426, #9430)"

Can you confirm if you were using the graphql-default naming convention? That might explain this behaviour

[CuriousLearner]

@rikinsk Yes, we have graphql-default naming-convention.