tower120
tower120
Have no idea. :)) It is at least twice more in terms of CPU cycles/latency.
"wouldn't making the size a template argument be "the c++ way"? That way it should produce same code and still be typesafe etc" What do you mean? Rephrase please.
Will you accept PR with implementation of this "feature"?
Fine by me. Thou `LinearPOT` could still benefit from the very fact of compile-tine pot checks with `ilog2`.
As I said before I did try division on pot number in godbolt. It replaces divisions with bit shifts. The thing is: - This is not guaranteed behavior. - You...
In `SegVec` you basically have `Vec`, with element access equal to accessing `**T`. Accessing elements in the same segments, have performance surprisingly close to per-index accessing `Vec` elements, due to...
OK. I can implement `raw_table` + `raw_table_mut` and PR it - that's not problem. Wouldn't it be a breaking change? Maybe something like `raw_table_immut` instead?
So can I do it now? Will you review/accept that PR?
OK - sure. I'm starting the implementation.
I have the same problem. Tried version 0.2.0 and 0.2.5. Windows version from website. I can open https://lapce.github.io/plugins.json from my browser. Also, I can't click on "Plugin settings" in Settings....