Stream: t-compiler/rust-analyzer

Topic: Maybe we should go to GHIR


Mr Smeet (Aug 15 2020 at 09:04, on Zulip):

I find reasonable up lvl of abstraction and use graph DSL like Cypher for directly accessing to graphs inside code. So what do you think about it?

Mr Smeet (Aug 15 2020 at 09:11, on Zulip):

If there are no objections or comments, immediately after the current task, I will move on to this task. Or maybe have sense use whole graph query language

matklad (Aug 15 2020 at 16:09, on Zulip):

I agree that we need some more structured way to represent Rust code for an IDE

matklad (Aug 15 2020 at 16:10, on Zulip):

https://github.com/GabrielDosReis/ipr seems like a good model to learn from

matklad (Aug 15 2020 at 16:11, on Zulip):

However, I think the main question is not the "shape" of model (graph, sea of objects, table of facts), but, rather, what are the entities that we are modeling.

matklad (Aug 15 2020 at 16:12, on Zulip):

As in, I have a very fuzzy boundaries in my head beween ide, hir, and hir_xxx crates. Without making those boundaries crisp, it doesn't really matter what specific formalism we use to describe things.

Last update: Jul 29 2021 at 08:15UTC