testingUtilities::nearlyEqualDbl has an integer conversion bug. It accepts type int when an internal function (testingUtilities::ulpsDistanceDbl) return type int64_t and this can lead to truncation issues when large ULP errors are in play. I've fixed this issue in my fork and a PR will be in shortly along with another minor fix.
bcaddy
commented
1 year ago
testingUtilities::nearlyEqualDblhas an integer conversion bug. It accepts typeintwhen an internal function (testingUtilities::ulpsDistanceDbl) return typeint64_tand this can lead to truncation issues when large ULP errors are in play. I've fixed this issue in my fork and a PR will be in shortly along with another minor fix.