diff options
authorTor Lillqvist <>2016-10-06 16:25:28 +0300
committerTor Lillqvist <>2016-10-06 16:47:04 +0300
commit3456004fd5fce1bb2545e1eded708a2366ad73bf (patch)
parent321a0625bd153dfb522d7f9574d698f632ab7262 (diff)
Add FIXME comment about the OpenCL approxEqual() we define
Note that the OpenCL approxEqual() is used only in the implementation of the RATE function, which is not in the trusted subset anyway, so that won't actually be used. Change-Id: Ide30921b1e1f48abb68b5dc64bc2da2fa802038b
1 files changed, 6 insertions, 0 deletions
diff --git a/sc/source/core/opencl/opinlinefun_finacial.cxx b/sc/source/core/opencl/opinlinefun_finacial.cxx
index f062954d4115..63d8a775b9c9 100644
--- a/sc/source/core/opencl/opinlinefun_finacial.cxx
+++ b/sc/source/core/opencl/opinlinefun_finacial.cxx
@@ -10,6 +10,12 @@
#ifndef SC_OPENCL_OPINLINFUN_finacial
#define SC_OPENCL_OPINLINFUN_finacial
std::string approxEqualDecl="bool approxEqual(double a, double b);\n";
+// FIXME: this approxEqual() is identical to what the C++ rtl_math_approxEqual() used to be, but
+// that has been improved in the meantime, so probably this should be, too? OTOH, this is used only
+// to compare against 0.0, so could be made much simpler, and actually questionable whether it works
+// as intended anyway.
std::string approxEqual =
"bool approxEqual(double a, double b)\n"