chore: stop tracking files compiled by TS and advertise node@16 as the min supported runtime

[jgravois]

once we land #69 and this one, i'm in favor of publishing arc.js@1.0.0.

technically tests are only failing in 14.x because jest itself calls String.replaceAll(), but given that we're targeting ES2022, 16x (which end-of-lifed back in 08/2023) feels like a sensible baseline to me.

ref: https://www.reddit.com/r/typescript/comments/ygqhle/comment/iua3nx8/

[springmeyer]

Would you be up for doing the release (and doing a pre-release first like a v1.0.0-alpha to ensure it can be installed and works correctly)?

[springmeyer]

Looks good, thanks so much. Only question is whether the root ./arc.js might be depended upon by downstream users? That said I'm okay with v1.0.0 being breaking if it is clearly documented how to upgrade.

[jgravois]

Only question is whether the root ./arc.js might be depended upon by downstream users?

this PR is definitely backwards-compatible.

I'm okay with v1.0.0 being breaking if it is clearly documented how to upgrade.

if you're not opposed to breaking a few eggs 🍳, i'd argue that dropping the CJS build altogether is an idea that has merit. folks like @sindresorhus have championed native ESM packages for years.

i'd just want to confirm that turf's own build tooling and testing framework support pure ESM dependencies (it appears that they do), before proposing the change in a separate PR.

Would you be up for doing the release (and doing a pre-release first like a v1.0.0-alpha to ensure it can be installed and works correctly)?

for sure. i can publish 0.2.1 first if you're not opposed to me exploring the idea above ☝️. that would ensure that #69 makes it into the wild in a release that folks that still need CJS can use.

you may already know about this, but npm pack is also a really handy command to dry run before publishing an immutable version.

edit: haha. I'm just parroting something I already said in 2024.

[springmeyer]

Thanks! Yeah, go for it.