Chuanqi Xu

Let's track the issue in https://github.com/llvm/llvm-project/issues/64586.

According to the discussion in WG21, this is consistency with the design intention.

>> CLion collects module information from all .ixx, .cppm, and .mxx files in the project and doesn’t rely on the specific approach you use to build modules. This is not...

> > We can collect the potential module files (by its suffixes) in the project tree. > > Please don't use extensions to decide which files are modules. It is...

After taking some times to look at the code, I have some lower-level ideas: It looks possible to support C++20 modules in `BackgroundIndex`. 1. We need to introduce a new...

> Is there an opportunity to reuse this persisted BMI cache between the indexer and the preambles of open files? I am not pretty sure that I've understood the idea...

> Note, I was also thinking of persisting the BMIs between clangd invocations. It makes sense. I feel like it is OK > The indexer also triggers AST builds (for...

> I think that would work as part of the solution. Another part of the solution would be to ensure that the indexing of source > files only starts after...

> module building looks kinda like indexing I am not sure what you mean for indexing. Abstractly, do you mean a map from source files to the BMI? If yet,...

> - we can scan the whole project lexically > - we can have the build system provide an artifact like "modules.json", generated at e.g. cmake time > - we...