SebastianVerified account

You’re sure perf wins is due to tagged literals specifically, and not due to other createElement inefficiencies this plans to address? https://github.com/reactjs/rfcs/blob/createlement-rfc/text/0000-create-element-changes.md …

Very sure. I have a fairly lengthy analysis of the performance implications of this approach that I haven't yet managed to share publicly.

Looking forward to looking into an end-to-end benchmark that shows what you’re comparing to Thanks for sharing!

I’ll delete my direct reply since that was based on a misunderstood premise of what you were trying to do. I’d still appreciate some writeups and details, e.g. as React repo issue or RFC. If you have evidence it’s a better output format then collaborating might be helpful?

At least we could help verify edge cases and check for perf regressions on large projects.

I can try to put something together. It's not as easy to benchmark as loading things in a browser, since there is far too much variability in metrics when testing indirectly.

For sure. I presume you have a methodology for observing perf wins from this which you can share

Yes. I was working with @mathias who is obviously pretty knowledgeable in the parser performance space. d8 she'll ships with a pretty good set of parser analysis tools that make things relatively straightforward.