numpydoc_xref_aliases always triggers sphinx rebuild

[bashtage]

Sphinx 2.2 reports the reason why a rebuild is needed. In a pretty simple demo site I am seeing:

updating environment: [config changed ('numpydoc_xref_aliases')] 179 added, 0 changed, 0 removed

as part of the build. This suggests that something about numpydoc_xref_aliases makes sphinx think that it is changing. I have not defined this variable, and I have no intersphinx.

[bashtage]

I am using master,

In [1]: import numpydoc
The history saving thread hit an unexpected error (OperationalError('attempt to write a readonly database')).History will not be written to the database.

In [2]: numpydoc.__version__
Out[2]: '1.0.0.dev0'

#226 seems related.

[bashtage]

When I load the pickle I see

In [11]: pkl.config.numpydoc_xref_aliases
Out[11]: 
{'None': ':data:`python:None`',
 'bool': ':ref:`bool <python:bltin-boolean-values>`',
 'boolean': ':ref:`bool <python:bltin-boolean-values>`',
 'True': ':data:`python:True`',
 'False': ':data:`python:False`',
 'list': ':class:`python:list`',
 'tuple': ':class:`python:tuple`',
 'str': ':class:`python:str`',
 'string': ':class:`python:str`',
 'dict': ':class:`python:dict`',
 'float': ':class:`python:float`',
 'int': ':class:`python:int`',
 'callable': ':func:`python:callable`',
 'iterable': ':term:`python:iterable`',
 'sequence': ':term:`python:sequence`',
 'contextmanager': ':func:`python:contextlib.contextmanager`',
 'namedtuple': ':func:`python:collections.namedtuple`',
 'generator': ':term:`python:generator`',
 'array': 'numpy.ndarray',
 'ndarray': 'numpy.ndarray',
 'np.ndarray': 'numpy.ndarray',
 'array-like': ':term:`numpy:array_like`',
 'array_like': ':term:`numpy:array_like`',
 'scalar': ':ref:`scalar <numpy:arrays.scalars>`',
 'RandomState': 'numpy.random.RandomState',
 'np.random.RandomState': 'numpy.random.RandomState',
 'np.inf': ':data:`numpy.inf`',
 'np.nan': ':data:`numpy.nan`',
 'numpy': ':mod:`numpy`'}

which are the defaults. Shouldn't this be an empty dict since I'm not passing in any options?

[larsoner]

#226 should have fixed this as it no longer modifies numpydoc_xref_aliases directly. Are you certain you're testing on latest master?

[bashtage]

Completely. I did an uninstall and then pip install git+https:// to make sure. I also verified the 226 function is present. Perhaps something new in Sphinx 2.2, which is pretty fresh?

[larsoner]

I run sphinx master most of the time and do not see this problem.

Are you sure your sphinx-build command is using that numpydoc install, though? For example if you add a raise RuntimeError in the func from 226 does your build break?

[bashtage]

100%. I ran sphinx in a debug build and stopped it in the #226 function.

[bashtage]

I'm using autosummary, not that this should make a difference.

[larsoner]

I can't replicate on master with our tinybuild env:

larsoner@bunk:~/python/numpydoc/numpydoc/tests/tinybuild$ make

rm -rf _build/*
rm -rf generated/
sphinx-build -b html -d _build/doctrees . _build/html
Running Sphinx v3.0.0+/aa290a1
making output directory... done
loading intersphinx inventory from https://docs.python.org/3/objects.inv...
[autosummary] generating autosummary for: index.rst
[autosummary] generating autosummary for: /home/larsoner/python/numpydoc/numpydoc/tests/tinybuild/generated/numpydoc_test_module.MyClass.rst, /home/larsoner/python/numpydoc/numpydoc/tests/tinybuild/generated/numpydoc_test_module.my_function.rst
building [mo]: targets for 0 po files that are out of date
building [html]: targets for 1 source files that are out of date
updating environment: [new config] 3 added, 0 changed, 0 removed
reading sources... [100%] index                                                                                                                                                     
looking for now-outdated files... none found
pickling environment... done
checking consistency... done
preparing documents... done
writing output... [100%] index                                                                                                                                                      
generating indices...  genindex py-modindexdone
writing additional pages...  searchdone
copying static files... ... done
copying extra files... done
dumping search index in English (code: en)... done
dumping object inventory... done
build succeeded.

The HTML pages are in _build/html.

larsoner@bunk:~/python/numpydoc/numpydoc/tests/tinybuild$ make html
sphinx-build -b html -d _build/doctrees . _build/html
Running Sphinx v3.0.0+/aa290a1
loading pickled environment... done
[autosummary] generating autosummary for: generated/numpydoc_test_module.MyClass.rst, generated/numpydoc_test_module.my_function.rst, index.rst
building [mo]: targets for 0 po files that are out of date
building [html]: targets for 0 source files that are out of date
updating environment: 0 added, 0 changed, 0 removed
looking for now-outdated files... none found
no targets are out of date.
build succeeded.

The HTML pages are in _build/html.

Do you see the same thing? If so then there must be some difference between our tinybuld and your doc build that causes the problem.

[bashtage]

Sorry for the trouble. Turns out it was a convoluted case of PEBCAK.

[larsoner]

:) No problem, glad it works properly!