merge no conflicts

theano/compile/profilemode.py

@@ -409,7 +409,7 @@ class ProfileMode(Mode):
             else:
                 msg = ' '
             print '   %4.1f%%  %5.1f%%  %5.3fs  %5.3fs  %.2es %s %5d %2d %2d %s' % (f, ftot, t, tot, t/nb_call, msg, nb_call, nb_op, nb_apply, a)
-        print '   ... (remaining %i Ops account for %.2f%%(%.2fs) of the runtime)'\
+        print '   ... (remaining %i single Op account for %.2f%%(%.2fs) of the runtime)'\
                 %(max(0, len(sotimes)-n_ops_to_print),
                   sum(f for f, t, a, ci, nb_call, nb_op in sotimes[n_ops_to_print:]),
                   sum(t for f, t, a, ci, nb_call, nb_op in sotimes[n_ops_to_print:]))
@@ -449,7 +449,7 @@ class ProfileMode(Mode):
                 print '   %4.1f%%  %5.1f%%  %5.3fs  %5.3fs  %.2es %s %7.1f %5d %2d %s' % (f, ftot, t, tot, t/nb_call, msg, op_flops.get(a,-1), nb_call, nb_apply, a)
             else:
                 print '   %4.1f%%  %5.1f%%  %5.3fs  %5.3fs  %.2es %s %5d %2d %s' % (f, ftot, t, tot, t/nb_call, msg, nb_call, nb_apply, a)
-        print '   ... (remaining %i Apply account for %6.2f%%(%.2fs) of the runtime)'\
+        print '   ... (remaining %i Op account for %6.2f%%(%.2fs) of the runtime)'\
                 %(max(0, len(otimes)-n_ops_to_print),
                   sum(f for f, t, a, ci, nb_call, nb_op in otimes[n_ops_to_print:]),
                   sum(t for f, t, a, ci, nb_call, nb_op in otimes[n_ops_to_print:]))

theano/compile/profiling.py

@@ -275,7 +275,7 @@ class ProfileStats(object):
                     ' (did you forget to enable counters?)')
             return
 
-        print >> file, 'Thunks'
+        print >> file, 'Apply'
         print >> file, '------'
         #print >> file, '<% time> <cumulative %%> <apply time> <cumulative seconds> <time per call> <nb_call> <Apply Op name>'
         # headers

theano/gof/cmodule.py

@@ -1041,7 +1041,8 @@ class ModuleCache(object):
         cached modules regardless of their age.
 
         :param clear_base_files: If True, then delete base directories
-        'cuda_ndarray', 'cutils_ext' and 'lazylinker_ext' if they are present.
+        'cuda_ndarray', 'cutils_ext', 'lazylinker_ext' and 'scan_perform'
+        if they are present.
         If False, those directories are left intact.
 
         :param delete_if_problem: See help of refresh() method.
@@ -1059,7 +1060,7 @@ class ModuleCache(object):
 
     def clear_base_files(self):
         """
-        Remove base directories 'cuda_ndarray', 'cutils_ext' and 'lazylinker_ext' if present.
+        Remove base directories 'cuda_ndarray', 'cutils_ext', 'lazylinker_ext' and 'scan_perform' if present.
 
         Note that we do not delete them outright because it may not work on
         some systems due to these modules being currently in use. Instead we
@@ -1068,7 +1069,7 @@ class ModuleCache(object):
         """
         compilelock.get_lock()
         try:
-            for base_dir in ('cuda_ndarray', 'cutils_ext', 'lazylinker_ext'):
+            for base_dir in ('cuda_ndarray', 'cutils_ext', 'lazylinker_ext', 'scan_perform'):
                 to_delete = os.path.join(self.dirname, base_dir + '.delete.me')
                 if os.path.isdir(to_delete):
                     try:

theano/tensor/basic.py

@@ -2143,18 +2143,23 @@ def switch(cond, ift, iff):
 @_scal_elemwise_with_nfunc('bitwise_and', 2, 1)
 def and_(a,b):
     """bitwise a & b"""
+bitwise_and = and_ # numpy name for it
 
 @_scal_elemwise_with_nfunc('bitwise_or', 2, 1)
 def or_(a,b):
     """bitwise a | b"""
+bitwise_or = or_ # numpy name for it
 
 @_scal_elemwise_with_nfunc('bitwise_xor', 2, 1)
 def xor(a,b):
     """bitwise a ^ b"""
+bitwise_xor = xor # numpy name for it
 
 @_scal_elemwise_with_nfunc('invert', 1, 1)
 def invert(a):
     """bitwise ~a"""
+bitwise_not = invert # numpy alias for it
+
 
 ##########################
 # Math
@@ -3152,14 +3157,13 @@ class Subtensor(Op):
         if not isinstance(node.inputs[0].type, TensorType):
             raise NotImplementedError()
         #
-        # two arrays are created:
+        # two arrays are created in C code:
         # is_slice: len == ndim, 0 means int, 1 means slice
         # subtensor_spec: len = n_ints + 3 * n_slices
         #
         fail = sub['fail']
-        init_cmds = []
+        init_cmds = [] # initialization for subtensor_spec
         is_slice = []
-        inplace = 1
         NONE_CODE = sys.maxint - 1
 
         pos = [0,1] #annoying version of global variable for init_entry