This reverts most of commit 85599f7582b8809be889f589fe4d1d95e09ec945; it turns out the ‘b’ flag changes behavior more drastically in Python 3.
Fixes SR-12328, rdar://problem/60230415.
GYB opens its inputs and outputs in text mode. This is a problem because the recommendation in Windows is to check everything out in LF mode to avoid breaking tests, so gybbing a file ends up converting LFs to CRLFs. That broke test/SILGen/magic_identifier_file_conflicting.swift.gyb.
Modify GYB to open its files in binary mode instead of text mode. This is a no-op everywhere but Windows.
Note that this change is incomplete: it’s somewhat difficult to switch stdin and stdout to binary mode in Python 2. I’ve added FIXME comments about this.
A number of gyb tests (gyb --test) were failing on Windows. It appears it was
introduced by commit d57b41e which altered the behaviour of line
directives to print paths with forward-slashes even on Windows.
This commit changes two areas of gyb.py:
1. Doctests use normalized paths where applicable.
2. Path normalization is done early in ParseContext initialization
instead of when formatting line directives in ExecutionContext.append_text.
Fixes: SR-9644
Most people don't know what this is for and set it to "", robbing themselves of usability benefits. I'm going to add more help text to line-directive to explain that.
When building on Windows, the filepath contains `\` as the path
separator. Substitute `/` for the path separator prior to replacement.
This allows clang to properly compile the source file. Windows supports
both as a path separator and this matches what clang does by default for
the filepaths when preprocessing. This allows gyb generated files to
build on Windows.
This patch reworks the line-directive command line argument for gyb to
take in a python-style format string with `line` and `file` variables
to be substituted.
Because utils/gyb is just a launcher, `doctest.testmod()` without
explicit module doesn't perform doctest of `gyb` module.
It used to perform doctest of `util/gyb`, the `__main__` module.
From the Python documentation:
* NotImplemented: "Special value which can be returned by the 'rich comparison'
special methods (__eq__(), __lt__(), and friends), to indicate that the
comparison is not implemented with respect to the other type."
* NotImplementedError: "This exception is derived from RuntimeError. In user
defined base classes, abstract methods should raise this exception when they
require derived classes to override the method."
* E101: indentation contains mixed spaces and tabs
* E111: indentation is not a multiple of four
* E128: continuation line under-indented for visual indent
* E302: expected 2 blank lines, found 1
* W191: indentation contains tabs
The repo contains roughly 80 Python scripts. "snake_case" naming is used for
local variables in all those scripts. This is the form recommended by the PEP 8
naming recommendations (Python Software Foundation) and typically associated
with idiomatic Python code.
However, in nine of the 80 scripts there were at least one instance of
"camelCase" naming prior to this commit.
This commit improves consistency in the Python code base by making sure that
these nine remaining files follow the variable naming convention used for
Python code in the project.
References:
* PEP 8: https://www.python.org/dev/peps/pep-0008/
* pep8-naming: https://pypi.python.org/pypi/pep8-naming
