Now the real thing:

Passing unicode doesn't work for properties but works in Python3 and for callables.

>>> from gi.repository import GUdev
>>> GUdev.Client(subsystems=[u'block'])
Traceback (most recent call last):

+ Trace 234686

>>>

Created attachment 297133 [details] [review]
Allow passing unicode lists to GStrv properties on Python 2

Review of attachment 297133 [details] [review]:

Overall this looks fine, just a quick question and some style nit picks. Thanks!

::: gi/pygi-value.c
@@ +941,3 @@
+    for (i = 0; i < argc; ++i) {
+        PyObject* item = PySequence_Fast_GET_ITEM (obj, i);
+        /* same as _pygi_marshal_from_py_utf8 */

Would it make sense to refactor and share code in both places?

@@ +944,3 @@
+        if(PyUnicode_Check (item)) {
+            PyObject *pystr_obj = PyUnicode_AsUTF8String (item);
+            if (!pystr_obj)

Please use 1TBS

@@ +954,3 @@
+        }
+#endif
+                goto error;

1TBS

@@ +963,3 @@
+
+error:
+    for (i = i - 1; i >= 0; i--)

1TBS

(In reply to Simon Feltman from comment #3)
> Review of attachment 297133 [details] [review] [review]:
> 
> Overall this looks fine, just a quick question and some style nit picks.
> Thanks!
> 
> ::: gi/pygi-value.c
> @@ +941,3 @@
> +    for (i = 0; i < argc; ++i) {
> +        PyObject* item = PySequence_Fast_GET_ITEM (obj, i);
> +        /* same as _pygi_marshal_from_py_utf8 */
> 
> Would it make sense to refactor and share code in both places?

The error/exception handling is a bit different, but yeah I can look into that. Any suggestion were to best put such a helper?

Created attachment 298332 [details] [review]
patch-v2

Updated patch only addressing the style issues.

Review of attachment 298332 [details] [review]:

Looks good. Please commit with rebase and minor style issue nit.

::: gi/pygi-value.c
@@ +942,3 @@
+        PyObject* item = PySequence_Fast_GET_ITEM (obj, i);
+        /* same as _pygi_marshal_from_py_utf8 */
+        if(PyUnicode_Check (item)) {

space after "if"

Thanks