Avoid using std::function in event dispatcher for performance about hazel

[Chlorie]

As @ilikenoodles2 benchmarked it, the speed improvement is definitely very small. Only 2.4 microseconds. So this is likely a micro optimization. "Premature optimization is the root of all evil".

But I don't think a 20%+ improvement is "micro optimization", since the event system is used all over the code.
Plus making it a template doesn't cost you anything, and it gives you better performance together with better IDE diagnostics. (Those red lines under your code is way better than matryoshka template substitution error after you hit the compile button.)
I agree that premature optimization is the root of all evil, but what optimization is premature depends on each person's opinion. If you're just spending time trying to do things right in my humble opinion it is not premature.

[LovelySanta]

In my opinion, an optimalisation must be measured in relative improvement instead of the overal time.

As @WhoseTheNerd said, 2.4us optimalisation might not sound alot, but on an operation that only took 11.9us to begin with, this is an optimalisation that is >20% as @Chlorie said.

I can totaly understand that for example #76 is a premature optimalisation, but this has a great impact for the functionality it provides:

• It is >20% faster to distribute events (keep in mind it will do this for each layer, for each event)
• Feature: passing wrong functor will cause compilor error

If it was up to me, I would suggest approve starting a PR to implement this...

[ilikenoodles2]

I'd like to note that when timing Application's OnEvent, I get an average time of 0.0119ms, whereas when templating the function, the average time is 0.0095ms. Definitely an improvement.

Edit: To clarify, this was with only one call to Dispatch

[CleverSource]

No one has even made a PR for this. OP should consider making a PR for this or someone else could, I'd be up for it though.

[WhoseTheNerd]

As @ilikenoodles2 benchmarked it, the speed improvement is definitely very small. Only 2.4 microseconds. So this is likely a micro optimization. "Premature optimization is the root of all evil".

[WhoseTheNerd]

It's been almost a month and PR is still not implemented. 🤔