There could be a load of other reasons, and I'm sure its nothing serious whatever the case, just seemed a bit odd
Seeing the Bun team do it, too, just leaves me with the cringe.
Anyway, happy to see these kinds of improvements. I appreciate the html import, but I'd rather it make it into web standards first so that we don't have the gross commonjs situation that came up around js imports. Also not sure how happy I am about it bundling react/ts in that html? I'll have to see how it plays out, but it seems like it would introduce complexity that I will have to opt out of, for the sake of "just works (with the modern frameworks that we know of, none of which I use)".
But hey, speed is always great! Shame they didn't include a deno comparison, but I'm guessing it's only barely slower, or there would be more noise about it?
Overall, none of this moves the needle for me (I avoid s3, I don't use postgres directly; only through libs, node compatibility is not relevant to me), but it does look like a lot of areas that I imagine they have been getting questions about from enterprise outfits, so it seems like they're focusing on where the money is, which is exactly what you would want from a dependency; especially one that is going to be as critical as your runtime.
