pkgsrc-Changes-HG archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

[pkgsrc/trunk]: pkgsrc/graphics/MesaLib Add comment about not using -ffast-ma...



details:   https://anonhg.NetBSD.org/pkgsrc/rev/384bdd442966
branches:  trunk
changeset: 542535:384bdd442966
user:      bjs <bjs%pkgsrc.org@localhost>
date:      Wed May 21 05:58:52 2008 +0000

description:
Add comment about not using -ffast-math pending further testing.
It seems to cause instability, at least for me on NetBSD/amd64 (4.99.63).

diffstat:

 graphics/MesaLib/hacks.mk |  8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diffs (22 lines):

diff -r b83097eaec8c -r 384bdd442966 graphics/MesaLib/hacks.mk
--- a/graphics/MesaLib/hacks.mk Wed May 21 04:12:22 2008 +0000
+++ b/graphics/MesaLib/hacks.mk Wed May 21 05:58:52 2008 +0000
@@ -1,4 +1,4 @@
-# $NetBSD: hacks.mk,v 1.7 2008/05/21 04:11:37 bjs Exp $
+# $NetBSD: hacks.mk,v 1.8 2008/05/21 05:58:52 bjs Exp $
 
 .if !defined(MESALIBS_HACKS_MK)
 MESALIBS_HACKS_MK=     # defined
@@ -17,8 +17,10 @@
 .include "../../mk/compiler.mk"
 ###
 ### Ordinarily, this is defined by the build.  However, in pkgsrc,
-### policy dictates that we do it here.  Let's trust a semi-modern
-### gcc with -ffast-math for now.
+### policy dictates that we do it here.  
+###
+### XXXbjs:  -ffast-math seems to cause instability, at least on amd64.
+###          I have disabled it pending further testing.
 ###
 .  if !empty(CC_VERSION:Mgcc-[34]*)
 PKG_HACKS+=            no-strict-aliasing



Home | Main Index | Thread Index | Old Index