Tomasz Pluskiewicz

I just encountered a little twist. Please consider this [snippet](https://runkit.com/embed/hr1kuyj9rfx6) and its output below ```js const $rdf = require('[email protected]') const N3 = require('n3') const ex = require('@rdfjs/[email protected]')('http://example.com/') const getStream =...

`extends` in type param does not work the way I think you expect it to. See this example in [TS playground](https://www.typescriptlang.org/play?#code/C4TwDgpgBASgJgMwLIHs4QPoEYtQLxQDkATogLRYB0WhAUKJLIquhgM7ACGx+RpCZDtzoNoAQQA2ElAHcIceMjTQCilphxQAPkyWshxWrQDGEzmzZQAcpwC28q8qgBvAL4mzFqACEzAOwBrR3QXd1NzSwBFAFdOOFCjUSgAZWiAIwArCGNgAB5aKABIQrUnCAAPYAg-OEtJaTkFZmU8Uv0uYgAaAsKAPl4bezhg6B1fTkCR7SgACjboCqqay3n2DoB+KBi4gC4oPwgANwhiAEpE8GhFZI7UzOzgXjusnNzVg16LxgBRcxBr9IvR4EZ4PXKEX5sf4AEQAYoRPkA) ```ts import { NamedNode, BlankNode, Quad } from 'rdf-js'...

Hey Matthieu > [...] closer to the WHATWG Readable Stream Not an authoritative answer, but "not likely" ™️ with the stream spec based in the node streams > I'd be...

You do have a point I could agree with introducing such a change. However, I think the RDF/JS spec should be by definition compatible with v2 of the stream implementation...

How is performance relevant exactly? RDF/JS is also a spec and not an implementation, right? The v3 of streams can be consumed as async iterator. So we can reflect that...

I still don't see your point. Node stream already implements async iteration since v11 (and experimental in v10). All RDF/JS Stream does is constrain the chunk to be `Quad` instances....

I'm not sure. I think we should try to be true to the spec. I see that `Dataset#addAll` indeed accepts an array as well as another dataset instance so we...

> I would give everyone of the GH team 'types' npm publish rights. The person who merges the PR is the one who publishes to npm. Just make this automatic...

:+1: to atlassian changesets but they will make deployment more difficult to automate. I have never really tried their proposed GitHub Actions workflow > since this repo contains the types...

Splitting would be useful if we'd want to version the types for each spec independently. Dependencies between them are not a problem and changesets handles that nicely. Otherwise the preferred...