Get full coverage in our bench marks project.

[glennawatson]

We should get a lot more coverage of different performance tests in our Benchmark project. So most major classes covered.

Extensions

• When Activated
• When Any
• When Any Value
• When Any Observable
• When Any Dynamic

Interactions

• Interaction
• Register Handlers
• Handle
• Interaction Context

Observable Properties

• Reactive Object
• Observable As Property Helper

Reactive Commands

• Create
• Create From Task
• Create From Observable
• Create Combined
• Invoke

[Kavignon]

Lately, I’ve been interested by everything performance so this could be a fun thing! are we talking of adding micro-benchmarks with a benchmark.NET? What’s the scope?

[glennawatson]

@Kavignon if you want to help out this is probably a good starter task. Essentially we want a base line of most of our existing code base so we aren't going backwards in terms of performance. There is not a platform runner at the moment for benchmark dot net for Android iOS so I would focus on core functionality first that didn't exist in the platforms folder and work from there.

[Kavignon]

You mention full coverage. How do I determine case coverage to make sure when I submit my PR, it makes sense with what you’d expect in the benchmark project?

[glennawatson]

Yeah coverage is a bit tricky since as far as I know there isn't any automated tools. I was just going to go through each file in the src/ReactiveUI folder and produce a benchmark for it. I wouldn't do the Platform folder

[Kavignon]

How about this idea; I’m going to close ReactiveUI and do a few benchmarks on my own. Then submit a PR for those and we can discuss on said PR what you would have like to see if I miss anything. When that portion is stable, I’ll get over the items I haven’t done beforehand. Sounds good?

[RLittlesII]

@Kavignon I'd recommend starting with one of the items above, there are 4 categories of benchmarks. I believe I have the ReactiveCommand started. Let me dust that branch off this weekend and submit a PR for it. Then you can just follow that PR for the remaining benchmarks. That way you don't invest too much effort. I'd recommend open a PR per category to keep it easier to manage. The bigger the PR the harder it is to review.

[Kavignon]

That sounds pretty good in fact! let me know here when your PR is ready!