David I. Lehn

One problem seems to be in `lib/documentLoaders/node.js`, at the bottom where `_fetch()` catches errors and sets a `cause: e`. That error is somehow dropped when it pops up the stack....

Seems the error chain is too deep and just prints out some deep details as `[Object]`. You can see the error in the test catch handler by using something like...

@gkellogg I think it depends how strict the checks should be. I think those regexes I added were derived from a spec. Hard to check how correct they even are!...

Looks like `rdfDoc` got cut off in https://github.com/digitalbazaar/jsonld.js/issues/474#issuecomment-1060193784. I'm not sure how interoperability is supposed to work between implementations as far as blank nodes go. They are kind of tricky....

You can pass in a custom `documentLoader` param that handles URLs in a custom way. For instance, returning static data or data loaded from local files.

Did this work in 1.0.1? Can you provide short test case doc and context? I think the code is out of date with some of the tests related to this...

The test skipping part looks ok. The `@version` check is maybe ok. @dlongley: do we want to add in small parts of 1.1 support like this to 0.4.x? Or should...

Thinking a bit more... the current way 1.1 tests are marked and skipped is problematic. Opened a test suite issue on this. Perhaps we can better skip all 1.1 related...

The 1.1 skipping is improved now since the test suite marks 1.1 tests with a specVersion property.

`@version` support added in 0.5.x here: https://github.com/digitalbazaar/jsonld.js/commit/f6a91c0642dbdfa29bf680085a8f5e6f5693e4b3