Current Behavior

When using a remote transaction in RDF4J and adding RDF data (e.g., Turtle) that requires a base URI, the base URI is serialized with angle brackets around it — for example:

<http://example.com/>

instead of the expected:

http://example.com/

On the remote end, the base URI is expected without < and >. This mismatch leads to invalid IRIs being constructed. If the data contains:

<foo> a <bar> .

The remote endpoint interprets <http://example.com/> literally, leading to corrupted IRIs:

<%3Chttp://example.com/foo> a <%3Chttp://example.com/bar> .

Expected Behavior

The base URI should be sent without < and > delimiters so the remote endpoint interprets it correctly and the created IRIs look like http://example.com/foo.

NOTE: I'm not sure what the protocol agreement is in this case but it should be the same on both client and server. Either both should use < and > or both should not use it.

Steps To Reproduce

1. Use RDF4J to open a remote transaction.
2. Add Turtle data with relative IRIs and provide the base URI externally.
3. Observe the serialized request payload — the base URI includes < and > delimiters.
4. On the remote server, inspect the created IRIs — they have %3C prepended (< but URL-encoded).

Code:

var repository = new HTTPRepository("http://localhost:8080/repositories/test");
try (var conn = repository.getConnection()) {
    conn.begin();
    conn.add(new StringReader( "<foo> a <bar> ."),
            "http://example.com/",
            RDFFormat.TURTLE);
    conn.commit();
    try (var statements = conn.getStatements(null, null, null, false)) {
        Rio.write(statements, System.out, RDFFormat.TURTLE);
    }
}

Version

5.1.4

Are you interested in contributing a solution yourself?

None

Anything else?

No response