Ruben Verborgh

Mhm I still see the `[[...slug]]` bug (https://gitlab.com/vincenttunru/penny/-/issues/15), but the `localhost/:1` confused me too; that is not expected. Could you please show details of one of the 400 errors? (Go...

I second this. I had seen the bug occur when the query string was not removed in a HTTP/HTTPS context, but this is not the issue here.

> This probably won't be a quick-fix. We can webpack them out for now. Also the crypto stuff is very heavy; we can likely do with https://github.com/LinkedDataFragments/Client.js/blob/master/lib/browser/crypto.js

> This is extensively being used by sparqlee. Also for hashing though.

Triple patterns with multiple occurrences of the same variable are part of TPF. They can have the following forms in theory: - `?s ?p/ ?s` (plausible) - `?s ?s ?o/`...

Regarding encoding, I think that the `_` and `?` cases should be the same. Blank variables are scoped to the entire request (IRI), so their name matters as well. Agree...

Can't the parser just autoconvert non-named variables to named variables? After all, `?subject=` is a shortcut for `?subject=?x`.

Would this also be a problem if the assignment was done intelligently? Like: always generate a variable name that is not already used?

On the one hand, this seems to be a problem with the HDT library not correctly converting blank nodes to the corresponding Jena representation. On the other hand, the TPF...

We should be able to do the same as in the [JavaScript code](https://github.com/LinkedDataFragments/Server.js/blob/v1.1.6/lib/datasources/Datasource.js#L79-L84) by just changing [this function](https://github.com/LinkedDataFragments/Server.Java/blob/master/src/org/linkeddatafragments/datasource/HdtDataSource.java#L107-L113) (probably on the base class level even).