aboutsummaryrefslogtreecommitdiffhomepage
path: root/m4
diff options
context:
space:
mode:
authorRobin Haberkorn <rhaberkorn@fmsbw.de>2025-10-20 00:02:34 +0200
committerRobin Haberkorn <rhaberkorn@fmsbw.de>2025-10-20 00:02:34 +0200
commit243a3be9185412a5a29ba189c77569fa96c85e89 (patch)
treeb5aaa8ec1820262392676ad64474efd18982efe3 /m4
parentf223cfada825d87dcab393611f604b8014f2b172 (diff)
scintilla: use AX_PTRDIFF_ALIASES_INT to check whether ptrdiff_t aliases int
* Scintilla had build problems if ptrdiff_t doesn't alias int if it has the same storage size (e.g. on NetBSD 10 (armv6)). * Unfortunately, Neil Hodgson refused to merge my previous Scintilla patches that would have fixed the issue at the core. See https://groups.google.com/g/scintilla-interest/c/STAv6LgLyCo * He only agreed to introducing a separate flag to work around the issue (`-DPTRDIFF_DOESNT_ALIAS_INT`). * In order to continue to support all standard C/C++ conforming platforms we therefore need a compile-time check to check for the aliasability of ptrdiff_t and int.
Diffstat (limited to 'm4')
-rw-r--r--m4/ax_ptrdiff_aliases_int.m416
1 files changed, 16 insertions, 0 deletions
diff --git a/m4/ax_ptrdiff_aliases_int.m4 b/m4/ax_ptrdiff_aliases_int.m4
new file mode 100644
index 0000000..f7175cd
--- /dev/null
+++ b/m4/ax_ptrdiff_aliases_int.m4
@@ -0,0 +1,16 @@
+
+AC_DEFUN([AX_PTRDIFF_ALIASES_INT], [
+ AC_CACHE_CHECK([whether ptrdiff_t aliases int], [ax_cv_ptrdiff_aliases_int], [
+ AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[
+ @%:@include <stddef.h>
+ ]], [[
+ int a = 23;
+ ptrdiff_t *b = &a;
+ ]])],
+ [ax_cv_ptrdiff_aliases_int=yes],
+ [ax_cv_ptrdiff_aliases_int=no])
+ ])
+ AS_IF([test "x$ax_cv_ptrdiff_aliases_int" = "xyes"], [
+ AC_DEFINE([PTRDIFF_ALIASES_INT], [1], [Whether ptrdiff_t aliases int])
+ ])
+])dnl