Performance tuning for very large projects?

[sebhoss]

I have created https://github.com/metio/kube-custom-resources-rs which currently sits at around 1 million lines of code. It is a single workspace with around 400 crates in it. Running something like rust-analyzer analysis-stats . blows up my computer and never finishes.

Any tips or ideas on how to make this project more manageable? Buy moar hardware? Can RA be configured to go through this project crate-by-crate?

[ChayimFriedman2]

It's not clear to me what you mean by "crate-by-crate". r-a already doesn't analyze code you don't look at, if code you look at doesn't depend on it (although there is the exception of cache priming, which you can disable).

Keep in mind that analysis-stats is not the typical workflow for an IDE, as it analyzes the body of all functions (while in IDE, we only analyze functions you look at). But we know r-a can be slow, sometimes abysmally slow. And there is no magic fix. Although if you can capture a profile of what's slow (by setting the environment variable RA_PROFILE=*>100, as described in the guide), that could maybe help.