I'm facing almost the same issue:

Caught exn in request "textDocument/documentSymbol"
get-lexer: 'before-lang-line

[Error - 11:58:25 PM] Request textDocument/documentSymbol failed.
  Message: internal error in method "textDocument/documentSymbol"
  Code: -32603 

It shows this error every time I type or delete a character from code. Thanks for your help.

This issue got reported multiple times on the Magic Racket repo as well; I’ll link the respective issues and their authors below:

• Missing #lang angers magic-racket, gives less than helpful output Eugleo/magic-racket#22 by @jasonhemann
• internal error in method "textDocument/documentSymbol" Code: -32603 Eugleo/magic-racket#23 by @togna

If there’s any way I can help, let me know.

Thank you @Eugleo for the response, seem the solution is to add #lang racket in the second line based on links provided.

Yeah, but still the langserver probably should report that in some sensoble way, not just throw a random exception. That's the reason I'm reporting this, anyway.

I'm getting this error when doing the first exercise in beautiful racket.

I'm getting this when placing #lang racket on the second line

The read-accept-reader parameter controls this.

Thanks; I am still very new to Racket.

@aymanosman I am not sure this is the right issue for this — but I’m actually following up on this bug: Eugleo/magic-racket#24

It does not matter what line number the Lang declaration is on: “#lang br/quicklang” results in that error.

Should I open up a new issue?

I think you problem is different.

Forget what I said above.

What you have is an actual syntax error. You must be following the tutorial wrong.

Hi everyone,

I think that the before-lang-line is best fixed by adding a case to text-document.rkt's function get-lexer that tests if the lexer is 'before-lang-line and just returns the racket-lexer again, or potentially just always gives back the racket-lexer if the return result is not a procedure?.

For @aymanosman 's error, this message is not coming from the language server, it is coming from br/quicklang, which is reading the syntax of its body without read-accept-reader turned on. This behavior is good and expected for br/quicklang, because it doesn't make sense for a file to have multiple #lang lines.