SebastianVerified account

With htm in the mode here the author writes JSX & React consumes a data structure that looks like JSX, so this doesn’t seems like something that React the library (same API) nor React the programming model (author writes same code) should care about. Optimization similar to DCE

I don’t say I care about whether people use HTM or not. Only about misleading marketing.

Your tweet sounds like you are assuming people would write tagged templates instead of JSX, where tagged templates are just an implementation detail of the optimizing compiler for JSX.

Ahh I misunderstood. My impression was that this tweet was promoting a one-time codebase conversion from JSX to HTM because it brings “sweet gainzz”. I see what Jason meant more now (although haven’t seen convincing evidence of “gainzz” yet).

It’s pretty counter-intuitive that HTM promotes compiling to createElement as an optimization but also promotes compiling JSX to HTM as an optimization.

yeah i am confused how template literals would be more efficient than, like, an AST with the static stuff lifted into the outer scope

I guess an *entirely* static htm block would be a single string and therefore efficient to cache?

It's the JS parse time. A tagged template can be parsed extremely quickly and most engines have specific optimizations in place to take advantage of this. A sufficiently fast runtime parser for the template's DSL can fall within the gap left by this improvement.