consult-lsp--file-symbols-candidates: No symbols

[Peter-Chou]

when I run consult-lsp-file-symbols, it showed no symbols. but I got right infos with consult-imenu.

consult-lsp-file-symbols has a big advantage to consult-imenu due to the narrowing mechanism. I really hope this command could run as expected. thx.

[gagbo]

Hello,

It looks like your LSP server either hasn't been started, or that it cannot deal with the current file. consult-imenu only uses imenu, which is basically a regexp-based feature. The "No symbols" message means the LSP server didn't give any response to the documentSymbols request.

Cheers

[Peter-Chou]

@gagbo, thanks for the quick reply. I used pyright python language server and it works well with consult-lsp-symbols.

Does consult-lsp-symbols also use documentSymbols request? If so, consult-lsp-file-symbols should also works.

[gagbo]

No, there are 2 calls,

• For consult-lsp-symbols it is workspace/symbols

  consult-lsp/consult-lsp.el

  Line 326 in 19606a0

  "workspace/symbol"

• For consult-lsp-file-symbols, it is documentSymbols

  consult-lsp/consult-lsp.el

  Line 497 in 19606a0

  (lsp-request "textDocument/documentSymbol"

I didn't check the protocol in a while to see if it changed, but theoretically that should not be a problem, so somehow pyright doesn't want to answer the documentSymbols call it seems

[sebastiansturm]

just today I noticed I'd get the same error (with a language server that definitely returned symbols), but in my case it was because consult-lsp--flatten-document-symbols doesn't respect lsp-use-plists. Can open a small PR the next few days though for my personal use I'd also have to change the default filtering because having all levels of symbols presented as a flat list seems overwhelming to me. (also, there was some error trying to invoke consult-lsp--file-symbols--transformer if symbols spanned multiple lines; in that case cend can be smaller than cbeg)

[gagbo]

All good points, it's indeed more likely that the heart of the issue is the use of lsp-use-plists, I forgot to check the code for compatibility, and the bug with cend < cbeg is also an issue because we try to use the same line for everything. Thanks for the report

[gagbo]

Both issues should be fixed now with the commit above ^. I also tagged it so eventually it'll also hit MELPA stable for the mad lads who trust it.

Don't hesitate to comment if that doesn't work