Fixing minor bugs in nbformat and saving.

* json separator is not ',' to avoid adding extra space at EOL.
* vs used throughout nbformat.current.
* Cell collapse is properly loaded from notebook.

IPython/frontend/html/notebook/static/js/codecell.js

@@ -824,6 +824,8 b' var IPython = (function (IPython) {'
             if (data.collapsed !== undefined) {
                 if (data.collapsed) {
                     this.collapse();
+                } else {
+                    this.expand();
                 };
             };
         };

IPython/frontend/html/notebook/static/js/notebook.js

@@ -1191,7 +1191,7 b' var IPython = (function (IPython) {'
         // We may want to move the name/id/nbformat logic inside toJSON?
         var data = this.toJSON();
         data.metadata.name = nbname;
-        data.nbformat = 2;
+        data.nbformat = 3;
         // We do the call with settings so we can set cache to false.
         var settings = {
             processData : false,

IPython/frontend/html/notebook/static/js/quickhelp.js

@@ -42,6 +42,7 b' var IPython = (function (IPython) {'
             {key: 'Ctrl-m y', help: 'code cell'},
             {key: 'Ctrl-m m', help: 'markdown cell'},
             {key: 'Ctrl-m t', help: 'plaintext cell'},
+            {key: 'Ctrl-m 1-6', help: 'heading 1-6 cell'},
             {key: 'Ctrl-m p', help: 'select previous'},
             {key: 'Ctrl-m n', help: 'select next'},
             {key: 'Ctrl-m i', help: 'interrupt kernel'},

IPython/nbformat/current.py

@@ -56,7 +56,7 b' def parse_py(s, **kwargs):'
     if m is not None:
         nbformat = int(m.group('nbformat'))
     else:
-        nbformat = 2
+        nbformat = 3
     return nbformat, s
 
 
@@ -65,16 +65,19 b' def reads_json(s, **kwargs):'
     nbformat, d = parse_json(s, **kwargs)
     if nbformat == 1:
         nb = v1.to_notebook_json(d, **kwargs)
-        nb = v2.convert_to_this_nbformat(nb, orig_version=1)
+        nb = v3.convert_to_this_nbformat(nb, orig_version=1)
     elif nbformat == 2:
         nb = v2.to_notebook_json(d, **kwargs)
+        nb = v3.convert_to_this_nbformat(nb, orig_version=2)
+    elif nbformat == 3:
+        nb = v3.to_notebook_json(d, **kwargs)
     else:
         raise NBFormatError('Unsupported JSON nbformat version: %i' % nbformat)
     return nb
 
 
 def writes_json(nb, **kwargs):
-    return v2.writes_json(nb, **kwargs)
+    return v3.writes_json(nb, **kwargs)
 
 
 def reads_py(s, **kwargs):
@@ -82,13 +85,15 b' def reads_py(s, **kwargs):'
     nbformat, s = parse_py(s, **kwargs)
     if nbformat == 2:
         nb = v2.to_notebook_py(s, **kwargs)
+    elif nbformat == 3:
+        nb = v3.to_notebook_py(s, **kwargs)
     else:
         raise NBFormatError('Unsupported PY nbformat version: %i' % nbformat)
     return nb
 
 
 def writes_py(nb, **kwargs):
-    return v2.writes_py(nb, **kwargs)
+    return v3.writes_py(nb, **kwargs)
 
 
 # High level API

IPython/nbformat/v3/nbjson.py

@@ -53,6 +53,7 b' class JSONWriter(NotebookWriter):'
         kwargs['cls'] = BytesEncoder
         kwargs['indent'] = 1
         kwargs['sort_keys'] = True
+        kwargs['separators'] = (',',': ')
         if kwargs.pop('split_lines', True):
             nb = split_lines(copy.deepcopy(nb))
         return json.dumps(nb, **kwargs)