better compliance with empty-path behavior recommended in RFC 3986 6.2.3 (and Twisted test suite)

TODO.md

@@ -1,5 +1,7 @@
 # hyperlink TODO
 
+* RFC 3986 6.2.3 (always slash on empty path and non-empty host, not
+  just when query string is present)
 * Polish logo
 * Optimize percent decoding
 * Get coverage up

hyperlink/_url.py

@@ -514,7 +514,10 @@ class URL(object):
         if port is None:
             port = SCHEME_PORT_MAP.get(scheme)
         if host and query and not path:
-            path = ()  # (u'',)
+            # per RFC 3986 6.2.3, "a URI that uses the generic syntax
+            # for authority with an empty path should be normalized to
+            # a path of '/'."
+            path = (u'',)
 
         # Now that we're done detecting whether they were passed, we can set
         # them to their defaults:
@@ -1049,7 +1052,7 @@ class URL(object):
         """
         scheme = self.scheme
         authority = self.authority(with_password)
-        path = u'/'.join(([u''] if (self.rooted and self.path) else [])
+        path = u'/'.join(([u''] if self.rooted else [])
                          + [_encode_path_part(segment, maximal=False)
                             for segment in self.path])
         query_string = u'&'.join(

hyperlink/test/test_url.py

@@ -97,7 +97,7 @@ ROUNDTRIP_TESTS = (
 
     'http://googlewebsite.com/e-shops.aspx',
     'http://example.com:8080/search?q=123&business=Nothing%20Special',
-    'http://hatnote.com:9000?arg=1&arg=2&arg=3',
+    'http://hatnote.com:9000/?arg=1&arg=2&arg=3',
     'https://xn--bcher-kva.ch',
     'http://xn--ggbla1c4e.xn--ngbc5azd/',
     'http://tools.ietf.org/html/rfc3986#section-3.4',
@@ -431,7 +431,7 @@ class TestURL(TestCase):
             URL.from_text("http://www.foo.com/a/nice/path/")
             .add(u"foo", u"bar").to_text())
         self.assertEqual(
-            "http://www.foo.com?foo=bar",
+            "http://www.foo.com/?foo=bar",
             URL(host=u"www.foo.com").add(u"foo", u"bar")
             .to_text())
         urlpath = URL.from_text(BASIC_URL)