incorporated review proposals

Signed-off-by: rstein <r.steinhagen@gsi.de>

algorithm/include/gnuradio-4.0/algorithm/dataset/DataSetEstimators.hpp

@@ -36,9 +36,8 @@ namespace estimators {
 template<typename T>
 [[nodiscard]] constexpr T computeCentreOfMass(const DataSet<T>& ds, std::size_t minIndex = 0UZ, std::size_t maxIndex = max_size_t, std::size_t signalIndex = 0, std::source_location location = std::source_location::current()) {
     maxIndex = detail::checkIndexRange(ds, minIndex, maxIndex, signalIndex, location);
-    ;
-    T com  = T(0);
-    T mass = T(0);
+    T com    = T(0);
+    T mass   = T(0);
 
     for (std::size_t i = minIndex; i < maxIndex; i++) {
         const T x     = getIndexValue(ds, dim::X, i);
@@ -66,10 +65,10 @@ template<std::ranges::random_access_range T, typename TValue = gr::meta::fundame
     for (upperLimit = index; upperLimit < data.size() && data[upperLimit] > maxHalf; upperLimit++) {
         // done in condition
     }
-    for (lowerLimit = index; lowerLimit >= 0 && data[lowerLimit] > maxHalf; lowerLimit--) {
+    for (lowerLimit = index; data[lowerLimit] > maxHalf; lowerLimit--) {
         // done in condition
     }
-    if (upperLimit >= data.size() || lowerLimit < 0) {
+    if (upperLimit >= data.size()) {
         return std::numeric_limits<TValue>::quiet_NaN();
     }
     return static_cast<TValue>(upperLimit - lowerLimit);
@@ -87,10 +86,10 @@ template<std::ranges::random_access_range T, typename TValue = typename T::value
     for (upperLimit = index; upperLimit < data.size() && data[upperLimit] > maxHalf; upperLimit++) {
         // done in condition
     }
-    for (lowerLimit = index; lowerLimit >= 0 && data[lowerLimit] > maxHalf; lowerLimit--) {
+    for (lowerLimit = index; data[lowerLimit] > maxHalf; lowerLimit--) {
         // done in condition
     }
-    if (upperLimit >= data.size() || lowerLimit < 0) {
+    if (upperLimit >= data.size()) {
         return std::numeric_limits<TValue>::quiet_NaN();
     }
     TValue lowerRefined = detail::linearInterpolate(value_t(lowerLimit), value_t(lowerLimit + 1), data[lowerLimit], data[lowerLimit + 1], maxHalf);
@@ -178,7 +177,7 @@ template<typename T, typename TValue = gr::meta::fundamental_base_value_type_t<T
     }
 
     std::vector<T> data(values.begin(), values.end()); // temporary mutable copy for in-place partitioning
-    auto           mid = data.begin() + data.size() / 2;
+    auto           mid = data.begin() + static_cast<std::ptrdiff_t>(data.size()) / 2;
     std::ranges::nth_element(data, mid);
 
     if ((data.size() & 1UZ) == 0UZ) {
@@ -520,7 +519,7 @@ std::optional<StepStartDetectionResult<T>> detectStepStart(D& ds, TValue thresho
 
     T thresholdCrossing = isRising ? initial + threshold * (max_val - initial) : initial - threshold * (initial - min_val);
     T triggerTime       = estimators::getZeroCrossing<MetaInfo::None>(ds, thresholdCrossing, indexMin, indexMax, signalIndex);
-    if (triggerTime == std::numeric_limits<TValue>::quiet_NaN()) {
+    if (std::isnan(gr::value(triggerTime))) {
         return std::nullopt; // failed to compute trigger time
     }
 
@@ -588,7 +587,7 @@ template<MetaInfo mode = MetaInfo::Apply, DataSetLike D, typename T = typename s
         prevRMS = localRMS;
     }
     T settlingLevel = estimators::getMean(dataSet, settlingTimeIdx, maxIndex + windowSize, signalIndex);
-    settlingTime += windowSize / 2UZ;
+    settlingTime += static_cast<int>(windowSize) / 2;
     if (settlingTime < 0) {
         return std::nullopt;
     }

algorithm/include/gnuradio-4.0/algorithm/dataset/DataSetMath.hpp

@@ -222,7 +222,7 @@ constexpr DataSet<T> applyMedian(const DataSet<T>& ds, std::size_t windowSize, s
         std::copy(signal.begin() + start, signal.begin() + end, medianWindow.begin());
 
         auto medianWindowView = std::span(medianWindow.data(), size);
-        auto midIter          = medianWindowView.begin() + (size / 2UZ);
+        auto midIter          = medianWindowView.begin() + static_cast<std::ptrdiff_t>(size / 2UZ);
         std::ranges::nth_element(medianWindowView, midIter);
 
         if ((size & 1UZ) == 0UZ) { // even-sized window -> take average around mid-point

algorithm/include/gnuradio-4.0/algorithm/dataset/DataSetTestFunctions.hpp

@@ -152,7 +152,7 @@ template<typename T, typename TValue = gr::meta::fundamental_base_value_type_t<T
     ds.timing_events.resize(1UZ);
     ds.extents = {1, static_cast<std::int32_t>(count)};
 
-    auto gauss = [](U x, U mu, U sigma) -> TValue { return std::exp(-std::pow((TValue(x) - TValue(mu)) / TValue(sigma), 2) / U(2)) / (TValue(sigma) * std::sqrt(TValue(2) * std::numbers::pi_v<TValue>)); };
+    auto gauss = [](U x, U mu, U sig) -> TValue { return std::exp(-std::pow((TValue(x) - TValue(mu)) / TValue(sig), 2) / U(2)) / (TValue(sig) * std::sqrt(TValue(2) * std::numbers::pi_v<TValue>)); };
 
     for (std::size_t i = 0; i < count; ++i) {
         ds.axis_values[0][i] = static_cast<TValue>(i);

algorithm/test/qa_DataSetEstimators.cpp

@@ -89,7 +89,11 @@ const boost::ut::suite<"DataSet<T> visual test functions"> _DataSetTestFcuntions
 
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wdouble-promotion"
+#pragma GCC diagnostic ignored "-Wfloat-conversion"
+#pragma GCC diagnostic ignored "-Wconversion"
+#ifdef __clang__
 #pragma GCC diagnostic ignored "-Wimplicit-float-conversion" // GCC & Clang: Loss or promotion of floating-point precision, disabled only for unit-tests
+#endif
 
 const boost::ut::suite<"DataSet<T> element-wise accessor"> _dataSetAccessors = [] {
     using namespace boost::ut;
@@ -495,7 +499,7 @@ const boost::ut::suite<"DataSet<T> math "> _dataSetMath = [] {
 
         auto ds_pulse_proto = generate::stepFunction<T>("pulse", 100, 20);
         for (std::size_t i = 50; i < 70; ++i) {
-            ds_pulse_proto.signal_values[i] = 1.0 - 0.01 * (i - 50); // gradual decrease at end-of-flat top
+            ds_pulse_proto.signal_values[i] = value_t(1.0) - value_t(0.01) * value_t(i - 50); // gradual decrease at end-of-flat top
         }
         for (std::size_t i = 70; i < 100; ++i) {
             ds_pulse_proto.signal_values[i] = 0.0; // final level
@@ -568,7 +572,7 @@ const boost::ut::suite<"DataSet<T> filter"> _dataSetFilter = [] {
 
         auto smoothed = filter::applyMovingAverage(ds, 3UZ);
 
-        std::vector<T> expected = {0.1, 0.2, 0.4, 0.6, 0.7};
+        std::vector<T> expected = {value_t(0.1), value_t(0.2), value_t(0.4), value_t(0.6), value_t(0.7)};
         for (std::size_t i = 0; i < expected.size(); ++i) {
             T val = smoothed.signal_values[i];
             expect(approx(val, expected[i], T(1e-3))) << fmt::format("smoothed value at index {}: expected {}, got {}", i, expected[i], val);

meta/include/gnuradio-4.0/meta/utils.hpp

@@ -26,10 +26,13 @@
 namespace gr {
 
 #pragma GCC diagnostic push // suppress unavoidable float/int/size_t conversion warnings
+#pragma GCC diagnostic ignored "-Wconversion"
+#ifdef __clang__
 #pragma GCC diagnostic ignored "-Wimplicit-int-float-conversion"
 #pragma GCC diagnostic ignored "-Wshorten-64-to-32"
 #pragma GCC diagnostic ignored "-Wimplicit-int-float-conversion"
 #pragma GCC diagnostic ignored "-Wimplicit-float-conversion"
+#endif
 
 using Size_t                            = std::uint32_t; // strict type definition in view of cross-platform/cross-compiler/cross-network portability similar to 'std::size_t' (N.B. which is not portable)
 inline constexpr Size_t      max_Size_t = std::numeric_limits<gr::Size_t>::max();