This isn't a good fix. The "path" of a URI isn't a filesystem path; it's a specific part of a URI (see RFC 3986). The path portion of a URI cannot be "absolute" or "relative". So it doesn't make sense to have this functionality in the URI class.
fn:resolve-uri() is not supposed to accept a filesystem path, so the test case path-to-uri.xq is not valid. If it DOES work on Unix as written, that's probably a different bug.
I am commenting specifically on the fix to bug 868329. I have nothing to say about the fix for bug 868325 except that it should probably be in a separate commit.
This isn't a good fix. The "path" of a URI isn't a filesystem path; it's a specific part of a URI (see RFC 3986). The path portion of a URI cannot be "absolute" or "relative". So it doesn't make sense to have this functionality in the URI class.
fn:resolve-uri() is not supposed to accept a filesystem path, so the test case path-to-uri.xq is not valid. If it DOES work on Unix as written, that's probably a different bug.
I am commenting specifically on the fix to bug 868329. I have nothing to say about the fix for bug 868325 except that it should probably be in a separate commit.