From b8e8e94a8b83f9f06d326ad6d9f42d70326d064c Mon Sep 17 00:00:00 2001
From: Lorenzo Moneta <Lorenzo.Moneta@cern.ch>
Date: Fri, 16 Sep 2016 09:48:58 +0200
Subject: [PATCH] Fix test for 32 bit machines. Add an extra tolerance in
test25
---
test/TFormulaParsingTests.h | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/test/TFormulaParsingTests.h b/test/TFormulaParsingTests.h
index 2fe3736c922..1ba19ce0b8c 100644
--- a/test/TFormulaParsingTests.h
+++ b/test/TFormulaParsingTests.h
@@ -3,6 +3,7 @@
#include "TF3.h"
#include "TFormula.h"
#include "TGraph.h"
+#include "TMath.h"
#include "Math/ChebyshevPol.h"
#include <limits>
@@ -425,7 +426,8 @@ bool test25() {
TF1 t1("t1","(x<190)?(-18.7813+(((2.49368+(10.3321/(x^0.881126)))*exp(-((x^-1.66603)/0.074916)))-(-17.5757*exp(-((x^-1464.26)/-7.94004e+06))))):(1.09984+(0.394544*exp(-(x/562.407))))");
double x = 2;
double y =(x<190)?(-18.7813+(((2.49368+(10.3321/(std::pow(x,0.881126))))*exp(-((std::pow(x,-1.66603))/0.074916)))-(-17.5757*exp(-((std::pow(x,-1464.26))/-7.94004e+06))))):(1.09984+(0.394544*exp(-(x/562.407))));
- ret = (t1.Eval(2) == y );
+ // this fails on 32 bits - put a tolerance
+ ret = TMath::AreEqualAbs(t1.Eval(2) , y , 1.E-8);
if (!ret) std::cout << "Error in test25 - t1 != y " << t1.Eval(2.) << " " << y << std::endl;
ok &= ret;
--
GitLab