Jesse Wright

Yes - got it working; thankyou! Working on the Comunica ESM migration in https://github.com/comunica/comunica/pull/1379

> I know there is a tendency to put types in dependencies but in my mind, types, just as typescript, are things we rely on at build time, not execution...

Would opt-in support for this behavior be an option? > The point is that, if you both use a sh:targetClass + sh:node, your node shape will be validated multiple times...

There is a closed PR https://github.com/RubenVerborgh/AsyncIterator/pull/81 which does it and has significant improvements - it was meant to be bundled into a bugger update that I never got around to.

Thanks for flagging! You can use https://share.hsforms.com/1ZMMECk5wT2qYhxouAW-zpQg1jy to sign up for a Solid mailing list. I'll update the link soon.

Thanks @bourgeoa

> This package used to have better performance than the fetch-based one. It would appear that the difference is negligible on average ```bash $ npx ts-node test.ts [class ActorHttpFetch extends...

So I had some variables wrong in the above experiment, turns out `ActorHttpFetch` is about 2x faster on node on average (c.f. https://github.com/jeswr/comunica-perf/actions/runs/9927282637/job/27422034531#step:6:11); which provides even more support to remove...

> https://www.npmjs.com/package/node-polyfill-webpack-plugin should resolve these issues. I've already tried that on [this branch](https://github.com/jeswr/comunica-perf/tree/webpack-attempt). > If you already tried this, I would prefer if we could keep actor-http-native, but somehow exclude...

The main reason not to would be due to the planned breaking changes around / removal of autostart discussed in #25, but given that they have been sitting there for...