CohenArthur

@tschwinge I've made the required changes and pushed them :) tested locally and it seems to work fine on a system where glibc < 2.34. tested with `--enable-languages=all` and `--enable-languages=rust`

I'm going to merge this PR because it has been merged upstream, sadly it introduce a regression because I'm too stupid for autotools and shell scripting in general, so I'll...

turns out that the bug/issue does not actually prevent the build from completing, it's just noisy but the build completes. I'll still push the fix ASAP

The attribute is used within libcore 1.49 so we'll need to handle it

I think we should keep this issue for properly implementing the fundamental attribute, but as of right now I believe it falls into the category of "only useful for invalid...

we're now storing the namespaces and inserting definitions properly - the issue is inserting them in the right Rib. `toplevel.insert_or_error_out(...)` inserts them in the topmost Rib, which is not correct....

but we can't know which namespace to get the Rib from before `Early` so we're hitting the same problem :( maybe FinalizeImports should be a fully fledged visitor as well?...

Storing all ribs gets messy and unwieldy. I guess having a `map` would make more sense, and we could just visit all `UseTree`s in `FinalizeImports` and get the import data...

last step is to declare imports in each namespace where they resolved properly. this is highlighted by name_resolution16.rs in the testsuite. the current implementation bails as soon as it finds...

Last hurdle comes from imports depending on each other, like so: ```rust mod foo { pub struct Foo; } use foo::Foo; use Foo as Fo; fn main() {} ``` this...