pbf VS connect-es

it's been about 3 years, but in Grafana at the time we were using something like ArrowJS + Arrow Flight + protobuf.js and then render the datasets into dashboards on Canvas, especially for streaming at ~20hz.

when i benchmarked the fastest lib to simply convert the protobuf decode (https://github.com/mapbox/pbf), it was 5x slower than native JSON parsing in browsers for dataframe-like structures (e.g. a few dozen 2k-long arrays of floats).

does it beat https://github.com/mapbox/pbf ?

at least in the frontend (without WASM), it depends.

i tested https://github.com/mapbox/pbf and while it was faster for deep/complex structs vs an unoptimized/repetative JSON blob, it was slower at shallow structs and flat arrays of stuff. if you spend a bit of time encode stuff as flat arrays to avoid mem alloc, JSON parsing wins by a lot since it goes through highly optimized C or assembly, while decoding protobuf in the JIT does not.

The runtime is taken from mapbox/pbf (with basic type definitions)

> However, it's important to note that browser support wasn't a primary focus in gRPC's design. This oversight necessitates an additional component, grpc-web, for browser accessibility. Furthermore, external services often have specific needs like caching and load balancing, which are not directly catered to by gRPC. Adopting gRPC for external services might require bespoke solutions to support these features.

The article should mention the Connect protocol for web-based Protobuf messaging:

https://connectrpc.com/

We can solve this with a service! Now there are many ways I could have started, but I decided to test out gRPC along the way. This was a mistake. I hoped for the best, but gRPC ended up not being a good choice for the web client. Why? you ask. The gRPC protocol works with all the bells and whistles of http when used server to server, but web clients are not as great. The Javascript client is dependent on http 2.0, and it requires a proxy like Envoy to work with a browser. What's more, I didn't love the structure of the generated web client. So through the process of working on this 'local first stack' I actually got sucked in to a big rabbit hole in making the rpc system work. I ended up going with Connect which is a tool that can create a service from a protobuf service definition, that also talks a simple http 1.1 protocol. What ultimately sold me on this solution as the best is that it also came with a very nice to use web client generation, and even plugs in to my favorite react http helper useQuery.

Our stack at Escape is written in multiple languages because each team has specific needs. We use TypeScript for its vibrant ecosystem, Python for cybersecurity research and Go for performance-sensitive tasks. To orchestrate cross-language task orchestration, we first developed a simple request-response protocol over HTTP, but it wasn't sustainable as the Escape codebase grew rapidly. We evaluated several technologies to replace our homegrown protocol, and two emerged as the most promising options: Connect and Temporal. The title gives it away, but the reason is far from obvious

protobuf messages we’ll configure (@bufbuild/connect-es)

try https://github.com/bufbuild/connect-web

As for better gRPC-web, you might want to look into connect-web https://github.com/bufbuild/connect-web

They already have! Connect (https://github.com/bufbuild/connect-web) is what you're looking for, as it's grpc-web compatible.

Ye, fwiw there is an example code size comparison here:

https://github.com/bufbuild/connect-web/blob/main/packages/c...

I'm sure someone will chime in on the implementation details, but hopefully others can give it a try with their projects!