Performance Considerations

This document serves as a brain-dump for any performance-related ideas or concerns.

In its experimental phase, Seamstress has yet to be properly profiled for the most common use-cases, so there's probably a lot of room for optimization.

Optimization: getComputedStyles() as a pure function

If we avoid callbacks (which may introduce side-effects), getComputedStyles() can be thought of as a pure function that operates on the results of getStyleState(), which itself should be a pure function of props, state, and context.

This means we can memoize getComputedStyles().

Why might this be awesome?

The benefits would be seen if we need to re-run render(), but all of our style calculations have already been computed.

In practice, this may only be beneficial for components with very large Component.style objects that need to be parsed and iterated over.

For memo lookups to be fast, we'd need a fast way to value-compare the results of getStyleState().

When testing out this idea, we can probably use Immutable.js to solve this problem, and hopefully replace the dependency with something lighter that serves our specific use-case. (Immutable adds around 60KB to bundle sizes).