Magnus

Is this PR still under consideration? This feature might be useful in PropCheck ([issue](https://github.com/alfert/propcheck/issues/30)) to check if a persisted counterexample is still valid, or if the generator used to produce...

See https://github.com/rusterlium/erlang_nif-sys/pull/39#issuecomment-582134181, can you migrate to `rustler_sys`?

See also https://github.com/rusterlium/rustler/issues/479#issuecomment-1244239121.

> I haven't looked into this much, but would making rust use a different target directory for each mix compilation mode be another option? The library which is loaded is...

Closing, I think we cannot do something here.

With #496, `:release` will be the default mode. This can be overridden manually in `use Rustler`.

Closing, feel free to reopen if this was not sufficiently answered.

> So seems like a non-breaking change to add it for the Atom type as well, I guess? Yes, I think that makes sense. @rusterlium/core opinions? I kinda worry that...

Interesting. I would not expect `inline` to make a real difference. Can we explain why we see a difference in the performance?

@kaaboaye do you think you could add a test for this? In `rustler_test`, any NIF could also load some data with that function. We could then test that this data...