File osp-float.diff of Package osp-toolkit

From: Jan Engelhardt <jengelh@inai.de>
Date: 2017-01-13 11:45:22.421378878 +0100

Due to a change between glibc-2.22 and 2.24, isnan now requires a
floating-point object and will reject integer objects. The fact
that an integer was passed seems like a bug when looking at the
other OSPM_ISNAN calls.

osptransapi.c: In function 'OSPPTransactionAccumulateTwoWayDelay':
osptransapi.c:986:13: error: non-floating-point argument in call to function '__builtin_isnan'
             OSPM_ISNAN(metrics.mean, tnisnan);

---
 src/osptransapi.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: TK-4_13_0-20161107/src/osptransapi.c
===================================================================
--- TK-4_13_0-20161107.orig/src/osptransapi.c
+++ TK-4_13_0-20161107/src/osptransapi.c
@@ -983,7 +983,7 @@ int OSPPTransactionAccumulateTwoWayDelay
             /* sample mean - have to cast Samples to a float to get some precision on the mean */
             mean = ((metrics.mean * currnumber) + (ospvMean * ospvSamples)) / (float)metrics.samples;
 
-            OSPM_ISNAN(metrics.mean, tnisnan);
+            OSPM_ISNAN(mean, tnisnan);
 
             if (tnisnan) {
                 errcode = OSPC_ERR_TRAN_INVALID_CALC;