* We detect if it could be used and output an information.
* We detect if its used in module mode and warn about not
being used there.
* We error out if it's used with "yes", but the file cannot be found.
* We already have plugins that use this information too, but were
using different code, e.g. for numpy.
* The real_prefix attribute is only there and most useful.
* This massively cleans up the backend scons file, where this was
noise in the backend scons code.
* This is in preparation for 3.10 which will fail to work that work.
* Extended numeric values to be hexadecimal.
* Using tuples over strings in testing.
* This uses nuitka_bool only where necessary, and bool otherwise.
* Solves a bunch of TODO of going via the C type objects
rather than hardcoding in helper functions.
* This also prepares more C type usage for other C types.
* Removed error checking from truth checking, done on the outside now.
* This does similar for lists, tuples, and sets.
* Also cache the decision if sets should be reversed, so we do not
use eval more often than necessary.
* Use helpers for lists and tuples that take references, cleaning
the generated code up a bit there too.
* Update check for conda envs to support Python > 3.6
* Checking if "Anaconda" is in sys.version string, lead to wrong results for Python ver. > 3.6.
* Update docstring of shallUseStaticLibPython
* Autoformat currently only sorts pylint disable comments, and does
not call black yet.
* This also adds a few doc strings that I failed to keep separate as a
commit, but who cares.
* At least MiniConda does this, potentially other Python variants too.
* Extension modules work, and as long as a shared library is nearby,
which it is for MiniConda, programs work too.
* Mostly for PyLint 2.0.0 and 1.9.2 compatiblity, avoiding local functions
for complex functions to avoid PyLint bug making a difference.
* Using more comparison chains as warned by PyLint.
* Need to blacklist files because they give false alarms, hopefully only
for 2.0.0
* Split up class re-formulation into 2 files, otherwise its easy to get
lost in the code of the other while looking at one.
* Programs won't work that way, but this solves the missing library
that is not really needed to be linked against.
* Manylinux should work with this for now.
* This avoids going by name, which apparently also changed for Python3.7
and made the old code stop working.
* Also in case of an error, make it raise a proper exception instead of
assertion no error merely.
* The support for pip is gone, not even old versions can be installed
anymore.
* Compiling this version yourself is no longer possible on modern Linux
due to OpeNSSL changes.
* Changing PyLint identifiers to symbolic instead of message ids, easier
to come up it and know.
* Make sure to error exit PyLint checking. This was regressed when making
it a tool.
* Some other PyLint cleanups.
* Added special node types for these, as they are to behave different
from coroutines and generators.
* Special code generation and compiled types for these too.
* Make a few coroutine functions accessible for re-use in asyncgen,
but generally they share surpringly little.
* Create all types for coroutines or asyncgens at once, instead of
one by one.
* This is nearly complete, most tests are working, but not all.
* Always output the best syntax error format, and do away with compatible
syntax errors. Instead make the comparison restricted to the message
of error only.
* This reduces tedious code and tried to mimic the various Python versions
errors for no good reasons.
* They added a few cases where column carets should be there, we add
those for improved mode, and detect the need for them.
* This should fix build errors in tests for current Debian unstable.
* The evaluation order of set literals is correct, but the values are
to be added in reverse on unfixed versions of Python.
* This adds a special node type for literals that only changes the
code generation and run time computation of set literal results,
for which the built-in set may not be the right thing to use.
* Switching from debug build to non-debug build wasn't properly
working.
* Test runner now detects the filename of debug CPython builds, and
check if it's there, and switches to it as necessary.
* This used to be static, but was recently changed in at least
"2.7.11+" in Debian, which is likely a snapshot without proper
version bump.
* Since this will probably also land in 3.5, lets detect this properly
at compile time, and turn it into a template. Then we don't have to
check different versions.
