Issue 39522: AST Unparser with unicode kinded constants

>>> from __future__ import annotations
>>> import ast
>>> x: u"a" = 3
>>> __annotations__["x"]
"'a'"
>>> ast.dump(ast.parse(__annotations__["x"])) == ast.dump(ast.parse('u"a"'))
False

I guess before touching constant part, we should wait for GH-17426 (afterward I can prepare a patch)

New changeset 43aeefa41915e4d3b0e68bbd4268c1c378a72dce by Batuhan Taşkaya in branch 'master':
bpo-39522: Use _PyUnicodeWriter_WriteStr instead of PyUnicode_AS_DATA (GH-19523)
https://github.com/python/cpython/commit/43aeefa41915e4d3b0e68bbd4268c1c378a72dce

New changeset 33986465bde2a2188537c4ef6cdb6055e348f31f by Pablo Galindo in branch 'master':
bpo-39522: Always initialise kind attribute in constant ast nodes (GH-19525)
https://github.com/python/cpython/commit/33986465bde2a2188537c4ef6cdb6055e348f31f

commit aade1cc453698e1bc48861b16955c2c2219ec521
Author: Batuhan Taşkaya <batuhanosmantaskaya@gmail.com>
Date:   Tue Apr 14 21:55:01 2020 +0300

    bpo-395222: Correctly unparse unicode prefix in ast_unparse.c (GH-19512)

Many buildbots failed. Example:

https://buildbot.python.org/all/#/builders/500/builds/288

======================================================================
FAIL: test_annotations (test.test_future.AnnotationsFutureTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.lto/build/Lib/test/test_future.py", line 156, in test_annotations
    eq("u'some_string'")
  File "/home/dje/cpython-buildarea/3.x.edelsohn-fedora-z.lto/build/Lib/test/test_future.py", line 150, in assertAnnotationEqual
    self.assertEqual(actual, expected)
AssertionError: "'some_string'" != "u'some_string'"
- 'some_string'
+ u'some_string'
? +


----------------------------------------------------------------------

I think that this was fixes by PR19525 and PR19523. For instance, the buildbot you mentioned is green when building those PRs:

https://buildbot.python.org/all/#/builders/500/builds/289

I checked and all current buildbot failures are related to the refleak in test_threading so I think this issue is fixed. I will close the issue, please reopen of I missed something or you would like to address something else :)