[pypy-commit] cffi release-1.4: hg merge default

[arigo]

Author: Armin Rigo <arigo at tunes.org>
Branch: release-1.4
Changeset: r2487:77c28957c976
Date: 2015-12-17 16:26 +0100
http://bitbucket.org/cffi/cffi/changeset/77c28957c976/

Log:	hg merge default

diff --git a/c/misc_thread_posix.h b/c/misc_thread_posix.h
--- a/c/misc_thread_posix.h
+++ b/c/misc_thread_posix.h
@@ -100,11 +100,11 @@
 #endif
 
 
-/* Seems that CPython 3.5 made our job harder.  Did not find out how
+/* Seems that CPython 3.5.1 made our job harder.  Did not find out how
    to do that without these hacks.  We can't use PyThreadState_GET(),
    because that calls PyThreadState_Get() which fails an assert if the
    result is NULL. */
-#if (PY_MAJOR_VERSION * 1000 + PY_MINOR_VERSION) >= 3005
+#ifndef _Py_atomic_load_relaxed   /* this was abruptly un-defined in 3.5.1 */
 void *volatile _PyThreadState_Current;
    /* XXX simple volatile access is assumed atomic */
 #  define _Py_atomic_load_relaxed(pp)  (*(pp))
diff --git a/doc/source/whatsnew.rst b/doc/source/whatsnew.rst
--- a/doc/source/whatsnew.rst
+++ b/doc/source/whatsnew.rst
@@ -3,6 +3,16 @@
 ======================
 
 
+v1.4.1
+======
+
+* Fix the compilation failure of cffi on CPython 3.5.0.  (3.5.1 works;
+  some detail changed that makes some underscore-starting macros
+  disappear from view of extension modules, and I worked around it,
+  thinking it changed in all 3.5 versions---but no: it was only in
+  3.5.1.)
+
+
 v1.4.0
 ======