Update TODO items

TODO

@@ -2,9 +2,10 @@ General fixes:
 =============
 
 - Instead of forming clusters immediately, only do it if we are reversing
-  the text.  We'd need a separate bit to indicate cluster start then
+  the text.  We'd need a separate bit to indicate cluster start then.
+  Right now, BTW, for non-native direction runs, we get the cluster wrong...
 
-- Fix tt kern on/off
+- Fix tt kern on/off and GPOS interaction
 
 - Remove fixed-size feature/lookup arrays in hb-ot-map
 
@@ -20,12 +21,26 @@ API issues to fix before 1.0:
 
 - get_table shouldn't return NULL
 
-- Use tags for hb_script_t?
+- Use tags for hb_script_t
 
 - Figure out how many .so objects, how to link, etc
 
 - Shall y axis progress downward instead of upward?
 
+- User-data support ala cairo
+
+- Real subclassing support for vfunc vectors
+
+- Investigate Mozilla's user-data issue, make sure it's addressed
+
+- Add hb-glib / hb-icu two-way script conversion API
+
+- Add hb-cairo glue
+
+- Fix blob, remove mutex, etc.
+
+- Add sanitize API (since may affect blob API)
+
 
 API to add (maybe after 1.0):
 ============================
@@ -34,21 +49,15 @@ API to add (maybe after 1.0):
 
 - Add hb_shape_plan()/hb_shape_execute()
 
-- Add sanitize API
-
 - Add query API for aalt-like features?
 
 - SFNT api? get_num_faces?
 
 - Full matrix instead of scale?
 
-- Add hb-glib / hb-icu script conversion API
-
 - Add segmentation API
 
-- Add hb-cairo
+- Add hb-fribidi?
-
-- Add hb-fribidi
 
 
 Build fixes:
@@ -62,7 +71,7 @@ Build fixes:
 Optimizations:
 =============
 
-- Avoid allocating blob objects internally for for_data() faces
+- Avoid allocating blob objects internally for for_data() faces?
 
 - Add caching layer to hb-ft