Protobuf.NET VS test-infra

This article requires that reader is familar with core concept of Akka.NET serialization (see https://getakka.net/articles/serialization/serialization.html) and ProtoBuf-Net library (see https://github.com/protobuf-net/protobuf-net).

The model used, a simple invoice which uses protobuf-net 1 NuGet package to store information in a binary file.

Not sure if it works with gRPC but I really like how Protobuf.NET uses attributes like other serializers instead of needing to write a .proto and generate (not-very-C#-friendly) classes from it. Well, until you need to interop with other languages.

I mentioned JSON, but there are many formats that are much more efficient. I can mention FlatBuffers, MessagePack and ProtoBuf. These are the ones I've used myself, and personally I'm most comfortable with MessagePack and ProtoBuf. I don't think the performance would be an issue if you had to choose between these three, it's mostly the API that is different.

The closest way would be using Grpc.Web + Protobuf.net. The overall experience is pretty close to WCF server + Client where you share a common interface and let the client and server just call through those.

Grpc Web with Blazor WASM is a really pleasant experience so far imo for my personal projects at home. You have strongly typed models and methods and you have choice on sharing the contract between client and server if you're using code first instead of proto IDL files (e.g. protobuf-net).

It's pretty straight-forward to follow the protobuf-net docs to serialize the report, but we should really GZIP the stream for sending to reduce bandwidth consumption and improve performance:

There are a number of Protobuf implementations for .NET Core, but I like protobuf-net as it's a nice, clean, Apache 2.0 Licensed implementation. It is also supported by protogen, a great online generator that will output protobuf-net classes ready for use (for its CSharp profile). If you open the latest schema from the link here, you can simply paste into the generator. NOTE: At the time of writing, [(js_preEncoded)=true] isn't supported by the generator, and can be removed from the proto schema.

> Protobuf-java is a little heavy [...] Just depending on the library adds 1.6MB and nearly 700 classes before you even generate your own message classes.

By comparison, protobuf-net [1] is about 260KB and 68 classes. Python's [2] is a 1MB package download (with source).

Why's the Java one so big?

[1] https://github.com/protobuf-net/protobuf-net

[2] https://pypi.org/project/protobuf

The trickiest part of decoding protobufs blind is how embedded messages and strings have the same tag type, but that's pretty easy to handle.

You can write a simple protobuf decoder in a couple hundred lines of code if you don't want to pull in the full protoc dependency: https://github.com/kubernetes/test-infra/blob/master/guberna... https://github.com/kubernetes/test-infra/blob/master/guberna...

Advanced tip: have a way for approved contributors to re-trigger CI. For example, Kubernetes, Kubeflow, and minikube use a tool called prow to run tests. Prow comes with a prow-bot that can accept different commands, automatically tagging issues or rerunning specific tests. Not for small projects, but this can alleviate maintainer headaches caused by trivial busy work.

It started as a small inside joke, but kinda just blew up. A bunch of us really loved UGG (which, if you haven’t played it, still highly recommend it even if you aren’t normally into video games), and as some of us have been known to do, take jokes a little too far/literal. A plug-in for the Kubernetes was built (https://github.com/kubernetes/test-infra/pull/14587) that posts goose pictures on command. Then there were some contests/activities put together at our contributor summit that were goose-themed.

Here's a minimal Python protobuf parser that can extract basic information in about 50 LOC.

https://github.com/kubernetes/test-infra/blob/master/guberna...