Maximilian Roos

> As I wrote here ([#2878 (comment)](https://github.com/PRQL/prql/pull/2878#issuecomment-1598399472)), I tried several queries. One of the three seems to compile correctly now, thanks! Thanks for trying. I hadn't realized that was with...

I did some quick diagnostics. Using this diff from the current `main` (similar to #2908 but uses an unstable feature): ```diff diff --git a/prql-compiler/prqlc/tests/project/Project.prql b/prql-compiler/prqlc/tests/project/Project.prql index 1fc6a97f..1b44377f 100644 --- a/prql-compiler/prqlc/tests/project/Project.prql...

One way of emulating this on a Mac — run `ulimit -s 1024`, and then `cargo test` in the same shell. At `512`, doctests fail (not sure why this is...

FYI this was fixed and then started reoccurring again after https://github.com/PRQL/prql/pull/3786. I'll try and open a PR with the test reenabled to track if it's still an issue

@aljazerzen I've been thinking about this some more. Is this a fundamental limitation with our compiling approach? Or is there some way of getting around it? > It looks like...

> I doubt this is a fundamantally bad approach - I've parsed enormous PRQL files with a relatively small stack size, so I don't think it applies to Linux too....

> 1MB of stack memory is not that much. IIUC Windows has a 1-2MB default stack size, which is why we get this problem on Windows or when reducing the...

> I've setup a Windows VM, downloaded git, rust toolchain and prql, compiled prqlc and parsed the test file. !! > I'm not able to reproduce this error. Does `cargo...

> * it happens in prqlc/prql-compiler/src/semantic/resolver/expr.rs#L60, during resolution, Interesting; this is a bit different from the [old location](https://github.com/PRQL/prql/issues/2857#issuecomment-1603190843) `{ fn: "::fold_expr", file: "/Users/maximilian/workspace/prql/prql-compiler/src/semantic/resolver.rs", line: 279 },` but possibly because the...

I thought the idea had lots of promised when we discussed it before! IIRC it would also manage "a PRQL project" — which to some extent has morphed into our...