remaining XXX and FIXME comments

[ev-br]

$ grep XXX -rn ./torch_np/
./torch_np/testing/utils.py:223:    # XXX: catch ValueError for subclasses of ndarray where iscomplex fail
./torch_np/testing/utils.py:398:    # XXX: catch ValueError for subclasses of ndarray where iscomplex fail
./torch_np/tests/test_function_base.py:32:    @pytest.mark.xfail(reason="XXX: arange(start=0, stop, step=1)")
./torch_np/tests/test_function_base.py:46:    @pytest.mark.xfail(reason="XXX: arange(..., dtype=bool)")
./torch_np/tests/test_ndarray_methods.py:49:        # XXX: move out to dedicated test(s)
./torch_np/tests/test_ndarray_methods.py:67:# XXX : order='C' / 'F'
./torch_np/tests/test_ndarray_methods.py:262:    @pytest.mark.skipif(reason="XXX: need ndarray.chooses")
./torch_np/tests/test_dtype.py:20:        marks=pytest.mark.xfail(reason="XXX: np.dtype() objects not supported"),
./torch_np/tests/test_dtype.py:30:            marks=pytest.mark.xfail(reason="XXX: namespaced dtypes not supported"),
./torch_np/tests/test_dtype.py:38:            marks=pytest.mark.xfail(reason="XXX: np.dtype() objects not supported"),
./torch_np/tests/test_basic.py:24:    # w.bincount,     # XXX: input dtypes
./torch_np/tests/test_ufuncs_basic.py:53:    @pytest.mark.skip(reason="XXX: unary ufuncs ignore the dtype=... parameter")
./torch_np/tests/test_ufuncs_basic.py:60:    @pytest.mark.skip(reason="XXX: unary ufuncs ignore the dtype=... parameter")
./torch_np/tests/test_ufuncs_basic.py:287:            with assert_raises((TypeError, RuntimeError)):  # XXX np.UFuncTypeError
./torch_np/tests/test_ufuncs_basic.py:323:            with assert_raises((TypeError, RuntimeError)):  # XXX np.UFuncTypeError
./torch_np/tests/test_ufuncs_basic.py:351:        with assert_raises((ValueError, RuntimeError)):  # XXX ValueError in numpy
./torch_np/tests/test_xps.py:113:    # dtypes. XXX: Remove the below sanity check and subsequent casting when
./torch_np/tests/test_xps.py:149:@pytest.mark.xfail(reason="XXX: support converting namespaced dtypes")
./torch_np/tests/numpy_tests/lib/test_shape_base_.py:623:    @pytest.mark.skip(reason="XXX: order='F' not implemented")
./torch_np/tests/numpy_tests/lib/test_shape_base_.py:632:    @pytest.mark.xfail(reason="XXX: noop in torch, while numpy raises")
./torch_np/tests/numpy_tests/lib/test_type_check.py:110:        # assert_(not isinstance(out, np.ndarray))  # XXX: 0D tensor, not scalar
./torch_np/tests/numpy_tests/lib/test_type_check.py:124:        # assert_(not isinstance(out, np.ndarray))  # XXX: 0D tensor, not scalar
./torch_np/tests/numpy_tests/lib/test_type_check.py:141:        # assert_(not isinstance(out, np.ndarray))  # XXX: 0D tensor, not scalar
./torch_np/tests/numpy_tests/lib/test_type_check.py:155:        # assert_(not isinstance(out, np.ndarray))  # XXX: 0D tensor, not scalar
./torch_np/tests/numpy_tests/lib/test_nanfunctions.py:429: ## XXX  _v.setflags(write=False)
./torch_np/tests/numpy_tests/core/test_scalar_methods.py:17:@pytest.mark.skip(reason='XXX: scalar.as_integer_ratio not implemented')
./torch_np/tests/numpy_tests/core/test_scalar_methods.py:123:@pytest.mark.skip(reason='XXX: implementation details of the type system differ')
./torch_np/tests/numpy_tests/core/test_shape_base.py:337:        # XXX: a single argument; relies on an ndarray being a sequence
./torch_np/tests/numpy_tests/core/test_dtype.py:67: # XXX: what is 'q'? on my 64-bit ubuntu maching it's int64, same as 'l'
./torch_np/tests/numpy_tests/core/test_dtype.py:137:   ## XXX: out dtypes do not have .descr
./torch_np/tests/numpy_tests/core/test_dtype.py:161:@pytest.mark.skip(reason="XXX: value-based promotions, we don't have.")
./torch_np/tests/numpy_tests/core/test_numeric.py:2097:        # XXX: test modified since there are array scalars
./torch_np/tests/numpy_tests/core/test_numeric.py:2160:        self.orders = {'C': 'c_contiguous'} # XXX: reeenable when implemented, 'F': 'f_contiguous'}
./torch_np/tests/numpy_tests/core/test_indexing.py:407:    @pytest.mark.xfail(reason="XXX: low-prio behaviour to support")
./torch_np/tests/numpy_tests/core/test_indexing.py:561:        reason="XXX: low-prio to support assigning complex values on floating arrays"
./torch_np/tests/numpy_tests/core/test_indexing.py:584:@pytest.mark.xfail(reason="XXX: requires broadcast() and broadcast_to()")
./torch_np/tests/numpy_tests/core/test_scalar_ctors.py:13:    @pytest.mark.xfail(reason='XXX: floats from strings')
./torch_np/tests/numpy_tests/core/test_scalar_ctors.py:21:    @pytest.mark.xfail(reason='XXX: floats from strings')
./torch_np/tests/numpy_tests/core/test_scalarmath.py:624:        # XXX: TypeError from numpy, RuntimeError from torch
./torch_np/tests/numpy_tests/core/test_scalarmath.py:643:        with assert_raises((TypeError, RuntimeError)):    # XXX: TypeError from numpy
./torch_np/tests/numpy_tests/core/test_multiarray.py:548:        # assert_(np.asfortranarray(d).flags.f_contiguous)   # XXX: f ordering
./torch_np/tests/numpy_tests/core/test_multiarray.py:1933:        # bounds check : XXX torch does not raise
./torch_np/tests/numpy_tests/core/test_multiarray.py:3845:    @pytest.mark.xfail(reason="XXX: take(..., mode='clip')")
./torch_np/tests/numpy_tests/core/test_multiarray.py:3852:    @pytest.mark.xfail(reason="XXX: take(..., mode='wrap')")
./torch_np/tests/numpy_tests/core/test_multiarray.py:3860:    @pytest.mark.xfail(reason="XXX: take(mode='wrap')")
./torch_np/tests/numpy_tests/core/test_multiarray.py:7450:    @pytest.mark.xfail(reason="XXX: place()")
./torch_np/tests/numpy_tests/core/test_multiarray.py:7465:    @pytest.mark.xfail(reason="XXX: putmask()")
./torch_np/tests/numpy_tests/core/test_multiarray.py:7487:    @pytest.mark.xfail(reason="XXX: ndarray.flat")
./torch_np/tests/numpy_tests/core/test_multiarray.py:7501:    @pytest.mark.skip(reason="XXX: npy_create_writebackifcopy()")
./torch_np/tests/numpy_tests/core/test_multiarray.py:7520:    @pytest.mark.skip(reason="XXX: npy_create_writebackifcopy()")
./torch_np/tests/numpy_tests/core/test_multiarray.py:7531:    @pytest.mark.skip(reason="XXX: npy_create_writebackifcopy()")
./torch_np/tests/test_reductions.py:73:        # XXX: numpy emits a warning on empty slice
./torch_np/tests/test_reductions.py:104:    @pytest.mark.xfail(reason="XXX: mean(..., where=...) not implemented")
./torch_np/_dtypes.py:168:    "longlong": int64,  # XXX: is this correct?
./torch_np/_dtypes.py:208:    "b": int8,  # XXX: srsly?
./torch_np/_reductions.py:224:        # XXX revisit when the pytorch version has pytorch/pytorch#95166
./torch_np/linalg.py:140:    # XXX: NumPy does this: https://github.com/numpy/numpy/blob/v1.24.0/numpy/linalg/linalg.py#L1744
./torch_np/_funcs_impl.py:154:    # XXX: in numpy 1.24 dstack does not have dtype and casting keywords
./torch_np/_funcs_impl.py:167:    # XXX: in numpy 1.24 column_stack does not have dtype and casting keywords
./torch_np/_funcs_impl.py:308:    # XXX: raises TypeError if start or stop are not scalars
./torch_np/_funcs_impl.py:359:        # XXX: this breaks if start is passed as a kwarg:
./torch_np/_funcs_impl.py:368:    # XXX: default values do not get normalized
./torch_np/_funcs_impl.py:443:    # XXX: fill_value broadcasts
./torch_np/_funcs_impl.py:679:    # XXX: semantic difference: np.flip returns a view, torch.flip copies
./torch_np/_funcs_impl.py:1063:    # XXX: scalar repeats; ArrayLikeOrScalar ?

and

$ grep FIXME -rn ./torch_np/
./torch_np/testing/__init__.py:17:# from .testing import assert_allclose    # FIXME
./torch_np/tests/numpy_tests/linalg/test_linalg.py:957:        #FIXME the 'e' dtype might work in future
./torch_np/tests/numpy_tests/lib/test_function_base.py:27:# FIXME: make from torch_np
./torch_np/tests/numpy_tests/lib/test_function_base.py:1173:#         np.array([0, 2**63, 0]),     # FIXME
./torch_np/tests/numpy_tests/lib/test_shape_base_.py:42:#  FIXME           (np.partition, np.argpartition, dict(kth=2)),
./torch_np/tests/numpy_tests/core/test_multiarray.py:80:# FIXME
./torch_np/tests/numpy_tests/core/test_multiarray.py:4705:        # FIXME:
./torch_np/_dtypes.py:161:# name aliases : FIXME (OS, bitness)
./torch_np/_reductions.py:400:    # FIXME(Mario) Doesn't np.quantile accept a tuple?

[lezcano]

Less than I expected. I reckon it should take us less than a day to triage all these into won't fix and fixable. Let's follow up after merge.