intel gave me trouble, can we avoid regex? is there an alternative to the line I commented out?

src/ArgumentParser.cpp

@@ -3,7 +3,9 @@
 #include <fstream>
 #include <iomanip>
 #include <iostream>
-#include <regex> // C++11
+#include <sstream>
+#include <limits>
+//#include <regex> // C++11
 
 #include "Cubism/ArgumentParser.h"
 
@@ -117,7 +119,9 @@ CommandlineParser::CommandlineParser(const int argc, char **argv)
                 // if the current value is numeric and (possibly) negative,
                 // do not interpret it as a key
                 const bool leadingDash = (argv[j][0] == '-');
-                const bool isNumeric = std::regex_match(argv[j], std::regex("(\\+|-)?[0-9]*(\\.[0-9]*)?((e|E)(\\+|-)?[0-9]*)?"));
+                const char c = argv[j][1];
+                const bool isNumeric = ((c>='0' && c<='9') || c==0)? true:false;
+                // const bool isNumeric = std::regex_match(argv[j], std::regex("(\\+|-)?[0-9]*(\\.[0-9]*)?((e|E)(\\+|-)?[0-9]*)?"));
                 if (leadingDash && !isNumeric)
                     break;
                 else

[kicici]

Here's a simple way to test if a string is a floating point number. A modified version of the stackoverflow answer:

#include <cmath>
#include <cstdlib>
// https://stackoverflow.com/questions/29169153/how-do-i-verify-a-string-is-valid-double-even-if-it-has-a-point-in-it
bool is_float(const char *s) {
    char *end = 0;
    double val = strtod(s, &end);
    return end != s && *end == '\0' && val != HUGE_VAL;
}

This one will also treat e.g. "-inf" and "-nan" as valid floating point values.

[fabianw]

What Intel compiler did you use? I can compile on panda with 2018.1.

Nevertheless, I will check what Ivica suggested and if it works as expected, I will remove the regex