Nathan Ridge

@lightmelodies Thanks for the very interesting pointers! The cache you are referring to is one I didn't know about. It's at a higher level and caches entire ASTs (preamble +...

@lixin-wei Here are my thoughts on an implementation approach suitable for experimentation: First, I do not think that extending the `ASTCache` which @lightmelodies modified is suitable for this purpose. The...

It may also be helpful to look at https://reviews.llvm.org/D112661, which also adds a cache of preambles of closed files, though for a slightly different / more ambitious purpose (reusing them...

> the lsp client will send notify when file is modified Not for closed files, right? Keep in mind that files can be modified outside of the editor, e.g. via...

Doesn't the **compiler** perform relative path expansion at the time it processes the `-I` argument?

I expect it would be interpreted relative to the directory in which the compile command is invoked, i.e. the `"directory"` field of the entries in `compile_commands.json`.

> Feel free to close this issue - or keep it open if you consider this to be useful functionality to be implemented in the future FWIW, I would consider...

> I think this is also basically a dupe of #137, right? Looks that way to me. > * we reuse the imported AST as the main file changes (I...

Ah, ok. This is not what I thought you meant in [this comment](https://github.com/clangd/clangd/issues/36#issuecomment-674021543); I thought you meant cases where the lambda's instantiated type can be deduced from the _declaration_ of...

What would you expect the behaviour to be if `foo()` calls the lambda with arguments of multiple different types? As in: ```c++ void foo(auto callback) { callback(5); callback("string"); } ```