Record and playback?

[alcarney]

It might be useful to use lsp-devtools record to capture an LSP session and use pytest-lsp to replay that session and make sure that the server behaves as expected

[eliericha]

Hello!

I tried using lsp-devtools to implement something similar to this issue. I wanted to record a replay file that includes protocol headers such that the session file can be piped directly as input to the server. The benefit of this is being able to pipe the file directly as input to a server for debugging outside the test environment.

I think it would be helpful if lsp-devtools record had an option to record raw messages at the transport level, including headers.

[alcarney]

I think it would be helpful if lsp-devtools record had an option to record raw messages at the transport level, including headers.

You could at one point pass a --capture-raw-output flag to lsp-devtools record and it would do that, but it looks like I broke it! 😅

I will have to look at fixing it at some point - and adding some tests around it!

[eliericha]

It would be great to fix --capture-raw-output :)

Bu ultimately I think I prefer to use a different method of recording, without lsp-devtools, in order to simplify the architecture and not have to manage an additional subprocess. Hence #195 :)

[qvalentin]

Playback would be really great for quick integration test creation.
Think of it like recording tests with playwright: https://playwright.dev/docs/codegen#record-a-new-test
You would just need a good assertion algorithm (similar to snapshot testing).