Bit of background - With our implementation, we need our deployment workflow to publish the avro schema to schema registry and are unable to use the GMS publish of the schema when it detects a difference. We recently were having an issue where the schema published by the workflow was different than what the GMS was publishing, so the GMS kept getting an error trying to publish the schema. The cause for the difference was special characters in the following files that the Github Actions runner was not generating correctly: MLModelFactors.pdl (’) MLFeatureDataType.pdl (—, 4 instances) EthicalConsiderations.pdl (–) We were able to get the runner to properly generate the characters by adding a file encoding flag to the Gradle build to use UTF-8 and are no longer having the issue, though it was a little confusing working through the issue. To prevent this from happening to others, can I create a PR replacing the special characters with standard ones?

Hey @faint-painting-38451 definitely feel free to open up a PR to do so! I’ll let @brainy-tent-14503 chime in with his thoughts as well

I’d be interested in the utf-8 flag, I am surprised that this is not the default, but might be something we need to look at too.

These are the steps we have in the workflow:

export GRADLE_OPTS="-Dfile.encoding=utf-8"

./gradlew :metadata-events:mxe-registration:build

Though from what we have seen, this environment variable is only needed in our Github Actions workflow. Locally it seems to be defaulting to utf-8 for myself and other team members without setting that environment variable, it was just the workflow that had the issue.