The sidebar symbols both offer goto-implementation (click on a function) and calltips (mouse-over tooltip). But in the sidebar you don't have a cursor position. I do not see how you can possibly implement the two functionalities using the LSP interfaces designed for that purposes, but using the intended interfaces - and only those - is @techee entire argument.

So to my understanding both sidebar features have to be implemented using the fallback.

I think you misunderstand the purpose of the goto-implementation, signature and documentSymbol calls:

goto-implementation: this is only meant to be used for control-click on an identifier in the editor (or using a keybinding for the symbol at the cursor position) - then, based on the identifier location in the current file, the server tries to find where that symbol is defined. clang is pretty good at that and really returns the right one, taking into account includes, type hierarchies etc. This is something that is extremely hard to do without full AST information.
signature - like goto-implementation, this is triggered e.g. by ( and the LSP server evaluates the identifier in front of (, its visibility from the precise location in the editor, and returns the best possible calltip. Again, extremely hard to do without AST.
autocompletion - like above

Now documentSymbol:

returns current document's symbols, their locations, the "detail" field I mentioned several times above and kind of the symbol. These are trivial to interpret so if one wants to display "detail" as calltip, one can do it (I do it with my plugin for the sidebar). Or go to symbol for this list is also trivial to implement. There's no need for server assistance here and LSP leaves these trivial things on clients. This is no "fallback" as you call it, you can do whatever you wish with the symbols obtained using this call.

But the fallback, as proposed in this PR, relies entirely on TM backed by ctags parsing.

No, this PR uses documentSymbol symbols for the sidebar (shows symbol name in the tree, shows "detail" as the calltip, generates tree based on children and assigns icon based on kind). I was just suggesting that if this PR is considered too big, I could sacrifice this LSP feature and could live with the TM implementation.

I can believe that you or @techee don't care about goto-implementation and calltips enough on the sidebar but I do.

But this works, I really don't know what you are talking about. Have you actually tried the bloody plugin? It's a few minutes of your time to install and test it - would really save hours of pointless discussions.

Hopefully it won't do nothing if I accidentally call a private function that is not available from the call site in question. I still want to jump to the implementation in that case.

Try it and see how clangd behaves. I would expect it to be reasonably fuzzy. When we mention "accurate", we mean that when you have hundreds of methods called get() in your C++ project, the goto will jump to the right one instead of showing a list of 100 gets.

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.