bug#37140: 27.0.50; calc gives "Wrong type argument: number-or-marker-p, inf" to many operations

["Mattias Engdegård" <mattiase@acm.org>]

[-- Attachment #1: Type: text/plain, Size: 191 bytes --]

ons 2019-08-28 klockan 11:00 +0200 skrev Andreas Schwab:

> -ffinite-math-only defines __FINITE_MATH_ONLY__=1, so this can be
> used as a test.

Thank you. Any objections to the following?



[-- Attachment #2: 0001-src-floatfns.c-Check-against-__FINITE_MATH_ONLY__-bu.patch --]
[-- Type: text/x-patch, Size: 1006 bytes --]

From 111493c77c8bfb4f3c55097beddae2e517153c24 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mattias=20Engdeg=C3=A5rd?= <mattiase@acm.org>
Date: Thu, 29 Aug 2019 12:29:47 +0200
Subject: [PATCH] * src/floatfns.c: Check against __FINITE_MATH_ONLY__
 (bug#37140)

---
 src/floatfns.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/src/floatfns.c b/src/floatfns.c
index 0a85df47de..49068bee77 100644
--- a/src/floatfns.c
+++ b/src/floatfns.c
@@ -48,6 +48,14 @@ Copyright (C) 1988, 1993-1994, 1999, 2001-2019 Free Software Foundation,
 
 #include <count-leading-zeros.h>
 
+/* Emacs needs proper handling of ±inf; correct printing as well as
+   important packages depend on it.  Make sure the user didn't specify
+   -ffinite-math-only, either directly or implicitly with -Ofast or
+   -ffast-math.  */
+#if defined __FINITE_MATH_ONLY__ && __FINITE_MATH_ONLY__
+ #error Emacs cannot be built with -ffinite-math-only
+#endif
+
 /* Check that X is a floating point number.  */
 
 static void
-- 
2.21.0