AVRO 4091: [C#] Allow previously parsed schemas to be referenced when parsing a schema

[rayokota]

What is the purpose of the change

Currently the C# Avro library does not have a way to parse a new schema that refers to named schemas that have been previously parsed.

The equivalent in the Java Avro library is to construct an instance of Schema.Parser and to use the same parser instance to parse the referenced schemas before the referring schema.

The change that is needed is to simply modify the following method in Schema.cs from internal to public:

internal static Schema Parse(string json, SchemaNames names, string encspace)

Without this change, tools that allow Avro schemas to reference one another, such as the Confluent Schema Registry, cannot interoperate well with C# applications.

Verifying this change

This change is a trivial rework / code cleanup without any test coverage.

Documentation

• Does this pull request introduce a new feature? (no)

[rayokota]

@martin-g @KalleOlaviNiemitalo @zcsizmadia could I kindly get a review? Thanks in advance!

[martin-g]

I am not familiar with C# and/or the Avro C# SDK ...
The change is trivial so I enabled the CI checks, but I cannot say whether the change is a good one or not. There are no new tests verifying the problem or the fix.

[KalleOlaviNiemitalo]

@rayokota, are you planning a change in https://github.com/confluentinc/confluent-kafka-dotnet that would make use of this public method? If you can post that as a draft PR there, it will be easier to understand what the benefit is.

[KalleOlaviNiemitalo]

If the method becomes public, then it should have argument null checks and trimming like in Parse(string json).

To avoid redundant checks and trimming when the caller has already done those, one could rename the existing internal static Schema Parse(string json, SchemaNames names, string encspace) to ParseInternal and add a public Parse wrapper with the checks and trimming.

[rayokota]

Thanks for the review @martin-g and @KalleOlaviNiemitalo ! I've incorporated your review feedback.

@KalleOlaviNiemitalo , I've renamed the existing method as ParseInternal as you suggested, and added a public method that calls it after performing argument null checks and json trimming.

Yes, I plan to use this method in https://github.com/confluentinc/confluent-kafka-dotnet , since Confluent Schema Registry supports Avro schemas that can reference other schemas by name. This allows schema reuse, as well as allows our customers to decompose very Avro large schemas into smaller ones.

As suggested, I've added a test showing how I intend to use this method. The test has a root schema that refers to a nested schema by name only. In Confluent Schema Registry we also store some metadata with the root schema to indicate which schema(s) it refers to. The Schema Registry client in https://github.com/confluentinc/confluent-kafka-dotnet will retrieve both schemas independently, and then use the method in this PR to reconstruct the root schema.

I've implemented this recomposition functionality for Schema Registry clients written in Java, go, Python, and Node.js, since the Avro libraries in those other languages also provide a means to compose schemas by name. This PR will allow C# to provide the same capability.

Please let me know if you would like me to add anything else to my explanation or this PR. Thanks in advance!