Update packaging to match main chart.js library

[stockiNail]

Fix #814

[kurkle]

Looks ok, but donwe really want to rename the umd files?

[stockiNail]

@kurkle I wen back to the original version, without using UMD in the dist files, as shared. Let me know if it's ok and if yes, let me know if you think we should publish new version 2.1.1 (bumping).

[kurkle]

Looks good to me, except it won't work in node/commonjs.

@etimberg is it ok for you to drop cjs support for plugins too? (main repo already did)

[kurkle]

If we want to support chart.js v3 on node/commonjs with these later annotation plugin versions, this needs to be umd build with .cjs extension

Something like I did in autocolors: https://github.com/kurkle/chartjs-plugin-autocolors/blob/main/package.json

[stockiNail]

Ok Let me update in next hours

[stockiNail]

@kurkle updated as suggested

[etimberg]

Looks good to me, except it won't work in node/commonjs.

@etimberg is it ok for you to drop cjs support for plugins too? (main repo already did)

Dropping is fine by me, but should definitely be a major version release

[stockiNail]

Looks good to me, except it won't work in node/commonjs.
@etimberg is it ok for you to drop cjs support for plugins too? (main repo already did)

Dropping is fine by me, but should definitely be a major version release

@etimberg ok that means all plugins/controllers in chartjs org must publish a major version for chartjs 4 compatibility, doesn’t it?

[etimberg]

Looks good to me, except it won't work in node/commonjs.
@etimberg is it ok for you to drop cjs support for plugins too? (main repo already did)

Dropping is fine by me, but should definitely be a major version release

@etimberg ok that means all plugins/controllers in chartjs org must publish a major version for chartjs 4 compatibility, doesn’t it?

I don't think it's mandatory if the plugin packaging doesn't change

[stockiNail]

@etimberg @kurkle with CHART.JS 4.1.1 this PR sounds to me useless because the issue should be fixed.

[kurkle]

There could be some benefits to from type: 'module', but maybe nothing worth the time before someone requests something explicit.