[BUG] std does not compute when using rolling window

[raykallen]

Describe the bug
cannot compute std over a rolling window in cuDF
NotImplementedError: <class 'cudf.dataframe.string.StringColumn'>

Steps/Code to reproduce bug

df = cudf.Series([1,1,1,8,8,8])
df.rolling(3).mean()    # <--- this works correctly
df.rolling(3).std()     # <--- this produces an error

Expected behavior
returns standard deviation over the rolling window

0	null
1	null
2	0
3	4.041452
4	4.041452
5	0

Environment overview (please complete the following information)

• Environment location: Docker
• Method of cuDF install: conda

Environment details
cudf version 0.9.0

[kkraus14]

This is still an issue as of the 0.14 release.

[kkraus14]

@shwina kicking over to you since I think you're the most familiar with rolling. We should more gracefully error, here's the current stacktrace I get with the above:

In [4]: df.rolling(3).std()
---------------------------------------------------------------------------
TypingError                               Traceback (most recent call last)
<ipython-input-4-30688c2fbe32> in <module>
----> 1 df.rolling(3).std()

~/git/cudf/python/cudf/cudf/core/window/rolling.py in __getattr__(self, key)
    178         if key == "obj":
    179             raise AttributeError()
--> 180         return self.obj[key].rolling(
    181             window=self.window,
    182             min_periods=self.min_periods,

~/git/cudf/python/cudf/cudf/core/series.py in __getitem__(self, arg)
    755             return self.iloc[arg]
    756         else:
--> 757             return self.loc[arg]
    758
    759     def __setitem__(self, key, value):

~/git/cudf/python/cudf/cudf/core/indexing.py in __getitem__(self, arg)
     92     def __getitem__(self, arg):
     93         try:
---> 94             arg = self._loc_to_iloc(arg)
     95         except (TypeError, KeyError, IndexError, ValueError):
     96             raise IndexError("Failed to convert index to appropirate row")

~/git/cudf/python/cudf/cudf/core/indexing.py in _loc_to_iloc(self, arg)
    109             try:
    110                 found_index = self._sr.index._values.find_first_value(
--> 111                     arg, closest=False
    112                 )
    113                 return found_index

~/git/cudf/python/cudf/cudf/core/column/numerical.py in find_first_value(self, value, closest)
    326         found = 0
    327         if len(self):
--> 328             found = cudautils.find_first(self.data_array_view, value)
    329         if found == -1 and self.is_monotonic and closest:
    330             if value < self.min():

~/git/cudf/python/cudf/cudf/utils/cudautils.py in find_first(arr, val, compare)
    200             else:
    201                 gpu_mark_found_int.forall(found.size)(
--> 202                     arr, val, found, arr.size
    203                 )
    204     from cudf.core.column import as_column

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/cuda/compiler.py in __call__(self, *args)
    265
    266         if isinstance(self.kernel, AutoJitCUDAKernel):
--> 267             kernel = self.kernel.specialize(*args)
    268         else:
    269             kernel = self.kernel

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/cuda/compiler.py in specialize(self, *args)
    767         argtypes = tuple(
    768             [self.typingctx.resolve_argument_type(a) for a in args])
--> 769         kernel = self.compile(argtypes)
    770         return kernel
    771

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/cuda/compiler.py in compile(self, sig)
    783                 self.targetoptions['link'] = ()
    784             kernel = compile_kernel(self.py_func, argtypes,
--> 785                                     **self.targetoptions)
    786             self.definitions[(cc, argtypes)] = kernel
    787             if self.bind:

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_lock.py in _acquire_compile_lock(*args, **kwargs)
     30         def _acquire_compile_lock(*args, **kwargs):
     31             with self:
---> 32                 return func(*args, **kwargs)
     33         return _acquire_compile_lock
     34

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/cuda/compiler.py in compile_kernel(pyfunc, args, link, debug, inline, fastmath, extensions, max_registers)
     55 def compile_kernel(pyfunc, args, link, debug=False, inline=False,
     56                    fastmath=False, extensions=[], max_registers=None):
---> 57     cres = compile_cuda(pyfunc, types.void, args, debug=debug, inline=inline)
     58     fname = cres.fndesc.llvm_func_name
     59     lib, kernel = cres.target_context.prepare_cuda_kernel(cres.library, fname,

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_lock.py in _acquire_compile_lock(*args, **kwargs)
     30         def _acquire_compile_lock(*args, **kwargs):
     31             with self:
---> 32                 return func(*args, **kwargs)
     33         return _acquire_compile_lock
     34

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/cuda/compiler.py in compile_cuda(pyfunc, return_type, args, debug, inline)
     44                                   return_type=return_type,
     45                                   flags=flags,
---> 46                                   locals={})
     47
     48     library = cres.library

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler.py in compile_extra(typingctx, targetctx, func, args, return_type, flags, locals, library, pipeline_class)
    566     pipeline = pipeline_class(typingctx, targetctx, library,
    567                               args, return_type, flags, locals)
--> 568     return pipeline.compile_extra(func)
    569
    570

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler.py in compile_extra(self, func)
    337         self.state.lifted = ()
    338         self.state.lifted_from = None
--> 339         return self._compile_bytecode()
    340
    341     def compile_ir(self, func_ir, lifted=(), lifted_from=None):

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler.py in _compile_bytecode(self)
    399         """
    400         assert self.state.func_ir is None
--> 401         return self._compile_core()
    402
    403     def _compile_ir(self):

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler.py in _compile_core(self)
    379                 self.state.status.fail_reason = e
    380                 if is_final_pipeline:
--> 381                     raise e
    382         else:
    383             raise CompilerError("All available pipelines exhausted")

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler.py in _compile_core(self)
    370             res = None
    371             try:
--> 372                 pm.run(self.state)
    373                 if self.state.cr is not None:
    374                     break

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_machinery.py in run(self, state)
    339                     (self.pipeline_name, pass_desc)
    340                 patched_exception = self._patch_error(msg, e)
--> 341                 raise patched_exception
    342
    343     def dependency_analysis(self):

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_machinery.py in run(self, state)
    330                 pass_inst = _pass_registry.get(pss).pass_inst
    331                 if isinstance(pass_inst, CompilerPass):
--> 332                     self._runPass(idx, pass_inst, state)
    333                 else:
    334                     raise BaseException("Legacy pass in use")

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_lock.py in _acquire_compile_lock(*args, **kwargs)
     30         def _acquire_compile_lock(*args, **kwargs):
     31             with self:
---> 32                 return func(*args, **kwargs)
     33         return _acquire_compile_lock
     34

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_machinery.py in _runPass(self, index, pss, internal_state)
    289             mutated |= check(pss.run_initialization, internal_state)
    290         with SimpleTimer() as pass_time:
--> 291             mutated |= check(pss.run_pass, internal_state)
    292         with SimpleTimer() as finalize_time:
    293             mutated |= check(pss.run_finalizer, internal_state)

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/compiler_machinery.py in check(func, compiler_state)
    262
    263         def check(func, compiler_state):
--> 264             mangled = func(compiler_state)
    265             if mangled not in (True, False):
    266                 msg = ("CompilerPass implementations should return True/False. "

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/typed_passes.py in run_pass(self, state)
     96                 state.return_type,
     97                 state.locals,
---> 98                 raise_errors=self._raise_errors)
     99             state.typemap = typemap
    100             if self._raise_errors:

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/typed_passes.py in type_inference_stage(typingctx, interp, args, return_type, locals, raise_errors)
     68
     69         infer.build_constraint()
---> 70         infer.propagate(raise_errors=raise_errors)
     71         typemap, restype, calltypes = infer.unify(raise_errors=raise_errors)
     72

~/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/typeinfer.py in propagate(self, raise_errors)
    984                                   if isinstance(e, ForceLiteralArg)]
    985                 if not force_lit_args:
--> 986                     raise errors[0]
    987                 else:
    988                     raise reduce(operator.or_, force_lit_args)

TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Invalid use of Function(<numba.cuda.compiler.DeviceFunctionTemplate object at 0x7f3d7b8e3908>) with argument(s) of type(s): (int64, unicode_type)
 * parameterized
In definition 0:
    LoweringError: Failed in nopython mode pipeline (step: nopython mode backend)
No definition for lowering <built-in method eq_impl of _dynfunc._Closure object at 0x7f3d380aa458>(int64, unicode_type) -> bool

File "../git/cudf/python/cudf/cudf/utils/utils.py", line 57:
def check_equals_int(a, b):
    return a == b
    ^

[1] During: lowering "$0.3 = a == b" at /home/nfs/kkraus/git/cudf/python/cudf/cudf/utils/utils.py (57)
    raised from /home/nfs/kkraus/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/utils.py:81
In definition 1:
    LoweringError: Failed in nopython mode pipeline (step: nopython mode backend)
No definition for lowering <built-in method eq_impl of _dynfunc._Closure object at 0x7f3d380aa458>(int64, unicode_type) -> bool

File "../git/cudf/python/cudf/cudf/utils/utils.py", line 57:
def check_equals_int(a, b):
    return a == b
    ^

[1] During: lowering "$0.3 = a == b" at /home/nfs/kkraus/git/cudf/python/cudf/cudf/utils/utils.py (57)
    raised from /home/nfs/kkraus/miniconda3/envs/cudf_dev/lib/python3.6/site-packages/numba/core/utils.py:81
This error is usually caused by passing an argument of a type that is unsupported by the named function.
[1] During: resolving callee type: Function(<numba.cuda.compiler.DeviceFunctionTemplate object at 0x7f3d7b8e3908>)
[2] During: typing of call at /home/nfs/kkraus/git/cudf/python/cudf/cudf/utils/cudautils.py (141)


File "../git/cudf/python/cudf/cudf/utils/cudautils.py", line 141:
def gpu_mark_found_int(arr, val, out, not_found):
    <source elided>
    if i < arr.size:
        if check_equals_int(arr[i], val):
        ^

[beckernick]

This is the original issue requesting rolling standard deviation. This request is now covered by #8695 which includes additional context and information, and will be closed by #8809 . I'm going to close this issue in favor of #8695 to consolidate further discussion.