Performance metrix

[DotNetNerd]

I think an important feature is how nothing performs, so I would like a performance metrix to ensure it provides industry leading performance. With nothing being essential to lots og applications, how it performs is critical.

[dzmitry-lahoda]

Nothing is class object created in GC memory. Measures are not needed - performance is unacceptable. More, it is wrongly named. GC garbage is not Nothing. So whole naming, which is part of implementation, should be swapped. New architecture and redesign is needed. I think author of this Nuget is junior developer who should go and learn from real master before making such bad things public.

[bartelink]

I'd like to see the benchmark verify this against an appropriate baseline - if this is going to be done, it should run like nothing else out there

[dzmitry-lahoda]

I have meditated for 2 days onto nothing about Nothing. I doubt we need to measure at all, because I can evaluate performance in my head, and baseline is unachievable.

First idea was about another Nothing, which is structure without anything. But it allows too many operations on it, as if it is Something. And I have great doubts it is really nothing.

Another idea was to have nothing as keyword in the language which will voided during compilation. But it is only one language support, while we want Nothing to be shared among all languages.

Last, we could change VM to represent nothing. But VM specification is immutable.

Finally, I conclude we cannot have Nothing, we cannot have base line, and there is no need to measure.

[dsyme]

Hawking radiation seems relevant here