Sebastian Fischer

So I conducted a simple benchmark for the range, and the results are basically identical. Maybe the compiler optimizes away the call into the closure? If this is the case,...

There is also [this discussion](https://users.rust-lang.org/t/closure-inlining/16690/4). > Yeah, I don't think that that's supported today, though I'm not sure. The compiler will already pretty aggressively inline, the attributes just let you...

I now ran a slightly more realistic benchmark, where an `Rc` is used. I also added the time that it takes for the vector creation to the benchmark for comparison....

I also checked whether this holds when we use the dedicated `.sum()` method on the `Iterator` and the result is also the same: ```rust fn range_closure(n: usize) -> i32 {...

one disadvantage of using trait objects here is the lack of inlining possibility. We should do some benchmarks on how much this diminishes the gains yielded by the closure /...

we could also hardcode some important closures in an enum for things like numeric operations and `is.na` so they can be inlined.

Maybe a simpler solution is to add a `TaskCallback` to the process that runs the browser that does something like `cat(".______before_next_browser_propt\n")`, that can be used in the `read_until_browse_prompt` function.

Have you mentioned these issues in the R-devel mailing list? Also, I would love to somehow contribute to this project, but am not too familiar with the tools that are...

> > Have you mentioned these issues in the R-devel mailing list? > > @sebffischer I haven't. I've never raised a feature proposal like that before, so I'm not exactly...

https://arxiv.org/abs/1905.02928