/** * Implement $(LINK2 https://digitalmars.com/articles/b62.html, Value Range Propagation). * * Copyright: Copyright (C) 1999-2022 by The D Language Foundation, All Rights Reserved * Authors: $(LINK2 https://www.digitalmars.com, Walter Bright) * License: $(LINK2 https://www.boost.org/LICENSE_1_0.txt, Boost License 1.0) * Source: $(LINK2 https://github.com/dlang/dmd/blob/master/src/dmd/intrange.d, _intrange.d) * Documentation: https://dlang.org/phobos/dmd_intrange.html * Coverage: https://codecov.io/gh/dlang/dmd/src/master/src/dmd/intrange.d */ module dmd.intrange; import core.stdc.stdio; import dmd.astenums; import dmd.mtype; import dmd.expression; import dmd.globals; private uinteger_t copySign(uinteger_t x, bool sign) { // return sign ? -x : x; return (x - cast(uinteger_t)sign) ^ -cast(uinteger_t)sign; } struct SignExtendedNumber { uinteger_t value; bool negative; static SignExtendedNumber fromInteger(uinteger_t value_) { return SignExtendedNumber(value_, value_ >> 63); } static SignExtendedNumber extreme(bool minimum) { return SignExtendedNumber(minimum - 1, minimum); } static SignExtendedNumber max() { return SignExtendedNumber(ulong.max, false); } static SignExtendedNumber min() { return SignExtendedNumber(0, true); } bool isMinimum() const { return negative && value == 0; } bool opEquals(const ref SignExtendedNumber a) const { return value == a.value && negative == a.negative; } int opCmp(const ref SignExtendedNumber a) const { if (negative != a.negative) { if (negative) return -1; else return 1; } if (value < a.value) return -1; else if (value > a.value) return 1; else return 0; } SignExtendedNumber opUnary(string op : "++")() { if (value != ulong.max) ++value; else if (negative) { value = 0; negative = false; } return this; } SignExtendedNumber opUnary(string op : "~")() const { if (~value == 0) return SignExtendedNumber(~value); else return SignExtendedNumber(~value, !negative); } SignExtendedNumber opUnary(string op : "-")() const { if (value == 0) return SignExtendedNumber(-cast(ulong)negative); else return SignExtendedNumber(-value, !negative); } SignExtendedNumber opBinary(string op : "&")(SignExtendedNumber rhs) const { return SignExtendedNumber(value & rhs.value); } SignExtendedNumber opBinary(string op : "|")(SignExtendedNumber rhs) { return SignExtendedNumber(value | rhs.value); } SignExtendedNumber opBinary(string op : "^")(SignExtendedNumber rhs) { return SignExtendedNumber(value ^ rhs.value); } SignExtendedNumber opBinary(string op : "+")(SignExtendedNumber rhs) { uinteger_t sum = value + rhs.value; bool carry = sum < value && sum < rhs.value; if (negative != rhs.negative) return SignExtendedNumber(sum, !carry); else if (negative) return SignExtendedNumber(carry ? sum : 0, true); else return SignExtendedNumber(carry ? ulong.max : sum, false); } SignExtendedNumber opBinary(string op : "-")(SignExtendedNumber rhs) { if (rhs.isMinimum()) return negative ? SignExtendedNumber(value, false) : max(); else return this + (-rhs); } SignExtendedNumber opBinary(string op : "*")(SignExtendedNumber rhs) { // perform *saturated* multiplication, otherwise we may get bogus ranges // like 0x10 * 0x10 == 0x100 == 0. /* Special handling for zeros: INT65_MIN * 0 = 0 INT65_MIN * + = INT65_MIN INT65_MIN * - = INT65_MAX 0 * anything = 0 */ if (value == 0) { if (!negative) return this; else if (rhs.negative) return max(); else return rhs.value == 0 ? rhs : this; } else if (rhs.value == 0) return rhs * this; // don't duplicate the symmetric case. SignExtendedNumber rv; // these are != 0 now surely. uinteger_t tAbs = copySign(value, negative); uinteger_t aAbs = copySign(rhs.value, rhs.negative); rv.negative = negative != rhs.negative; if (ulong.max / tAbs < aAbs) rv.value = rv.negative - 1; else rv.value = copySign(tAbs * aAbs, rv.negative); return rv; } SignExtendedNumber opBinary(string op : "/")(SignExtendedNumber rhs) { /* special handling for zeros: INT65_MIN / INT65_MIN = 1 anything / INT65_MIN = 0 + / 0 = INT65_MAX (eh?) - / 0 = INT65_MIN (eh?) */ if (rhs.value == 0) { if (rhs.negative) return SignExtendedNumber(value == 0 && negative); else return extreme(negative); } uinteger_t aAbs = copySign(rhs.value, rhs.negative); uinteger_t rvVal; if (!isMinimum()) rvVal = copySign(value, negative) / aAbs; // Special handling for INT65_MIN // if the denominator is not a power of 2, it is same as ulong.max / x. else if (aAbs & (aAbs - 1)) rvVal = ulong.max / aAbs; // otherwise, it's the same as reversing the bits of x. else { if (aAbs == 1) return extreme(!rhs.negative); rvVal = 1UL << 63; aAbs >>= 1; if (aAbs & 0xAAAAAAAAAAAAAAAAUL) rvVal >>= 1; if (aAbs & 0xCCCCCCCCCCCCCCCCUL) rvVal >>= 2; if (aAbs & 0xF0F0F0F0F0F0F0F0UL) rvVal >>= 4; if (aAbs & 0xFF00FF00FF00FF00UL) rvVal >>= 8; if (aAbs & 0xFFFF0000FFFF0000UL) rvVal >>= 16; if (aAbs & 0xFFFFFFFF00000000UL) rvVal >>= 32; } bool rvNeg = negative != rhs.negative; rvVal = copySign(rvVal, rvNeg); return SignExtendedNumber(rvVal, rvVal != 0 && rvNeg); } SignExtendedNumber opBinary(string op : "%")(SignExtendedNumber rhs) { if (rhs.value == 0) return !rhs.negative ? rhs : isMinimum() ? SignExtendedNumber(0) : this; uinteger_t aAbs = copySign(rhs.value, rhs.negative); uinteger_t rvVal; // a % b == sgn(a) * abs(a) % abs(b). if (!isMinimum()) rvVal = copySign(value, negative) % aAbs; // Special handling for INT65_MIN // if the denominator is not a power of 2, it is same as ulong.max % x + 1. else if (aAbs & (aAbs - 1)) rvVal = ulong.max % aAbs + 1; // otherwise, the modulus is trivially zero. else rvVal = 0; rvVal = copySign(rvVal, negative); return SignExtendedNumber(rvVal, rvVal != 0 && negative); } SignExtendedNumber opBinary(string op : "<<")(SignExtendedNumber rhs) { // assume left-shift the shift-amount is always unsigned. Thus negative // shifts will give huge result. if (value == 0) return this; else if (rhs.negative) return extreme(negative); uinteger_t v = copySign(value, negative); // compute base-2 log of 'v' to determine the maximum allowed bits to shift. // Ref: https://graphics.stanford.edu/~seander/bithacks.html#IntegerLog // Why is this a size_t? Looks like a bug. size_t r, s; r = (v > 0xFFFFFFFFUL) << 5; v >>= r; s = (v > 0xFFFFUL ) << 4; v >>= s; r |= s; s = (v > 0xFFUL ) << 3; v >>= s; r |= s; s = (v > 0xFUL ) << 2; v >>= s; r |= s; s = (v > 0x3UL ) << 1; v >>= s; r |= s; r |= (v >> 1); uinteger_t allowableShift = 63 - r; if (rhs.value > allowableShift) return extreme(negative); else return SignExtendedNumber(value << rhs.value, negative); } SignExtendedNumber opBinary(string op : ">>")(SignExtendedNumber rhs) { if (rhs.negative || rhs.value > 63) return negative ? SignExtendedNumber(-1, true) : SignExtendedNumber(0); else if (isMinimum()) return rhs.value == 0 ? this : SignExtendedNumber(-1UL << (64 - rhs.value), true); uinteger_t x = value ^ -cast(int)negative; x >>= rhs.value; return SignExtendedNumber(x ^ -cast(int)negative, negative); } SignExtendedNumber opBinary(string op : "^^")(SignExtendedNumber rhs) { // Not yet implemented assert(0); } } struct IntRange { SignExtendedNumber imin, imax; this(IntRange another) { imin = another.imin; imax = another.imax; } this(SignExtendedNumber a) { imin = a; imax = a; } this(SignExtendedNumber lower, SignExtendedNumber upper) { imin = lower; imax = upper; } static IntRange fromType(Type type) { return fromType(type, type.isunsigned()); } static IntRange fromType(Type type, bool isUnsigned) { if (!type.isintegral() || type.toBasetype().ty == Tvector) return widest(); uinteger_t mask = type.sizemask(); auto lower = SignExtendedNumber(0); auto upper = SignExtendedNumber(mask); if (type.toBasetype().ty == Tdchar) upper.value = 0x10FFFFUL; else if (!isUnsigned) { lower.value = ~(mask >> 1); lower.negative = true; upper.value = (mask >> 1); } return IntRange(lower, upper); } static IntRange fromNumbers2(SignExtendedNumber* numbers) { if (numbers[0] < numbers[1]) return IntRange(numbers[0], numbers[1]); else return IntRange(numbers[1], numbers[0]); } static IntRange fromNumbers4(SignExtendedNumber* numbers) { IntRange ab = fromNumbers2(numbers); IntRange cd = fromNumbers2(numbers + 2); if (cd.imin < ab.imin) ab.imin = cd.imin; if (cd.imax > ab.imax) ab.imax = cd.imax; return ab; } static IntRange widest() { return IntRange(SignExtendedNumber.min(), SignExtendedNumber.max()); } IntRange castSigned(uinteger_t mask) { // .... 0x1e7f ] [0x1e80 .. 0x1f7f] [0x1f80 .. 0x7f] [0x80 .. 0x17f] [0x180 .... // // regular signed type. We use a technique similar to the unsigned version, // but the chunk has to be offset by 1/2 of the range. uinteger_t halfChunkMask = mask >> 1; uinteger_t minHalfChunk = imin.value & ~halfChunkMask; uinteger_t maxHalfChunk = imax.value & ~halfChunkMask; int minHalfChunkNegativity = imin.negative; // 1 = neg, 0 = nonneg, -1 = chunk containing ::max int maxHalfChunkNegativity = imax.negative; if (minHalfChunk & mask) { minHalfChunk += halfChunkMask + 1; if (minHalfChunk == 0) --minHalfChunkNegativity; } if (maxHalfChunk & mask) { maxHalfChunk += halfChunkMask + 1; if (maxHalfChunk == 0) --maxHalfChunkNegativity; } if (minHalfChunk == maxHalfChunk && minHalfChunkNegativity == maxHalfChunkNegativity) { imin.value &= mask; imax.value &= mask; // sign extend if necessary. imin.negative = (imin.value & ~halfChunkMask) != 0; imax.negative = (imax.value & ~halfChunkMask) != 0; halfChunkMask += 1; imin.value = (imin.value ^ halfChunkMask) - halfChunkMask; imax.value = (imax.value ^ halfChunkMask) - halfChunkMask; } else { imin = SignExtendedNumber(~halfChunkMask, true); imax = SignExtendedNumber(halfChunkMask, false); } return this; } IntRange castUnsigned(uinteger_t mask) { // .... 0x1eff ] [0x1f00 .. 0x1fff] [0 .. 0xff] [0x100 .. 0x1ff] [0x200 .... // // regular unsigned type. We just need to see if ir steps across the // boundary of validRange. If yes, ir will represent the whole validRange, // otherwise, we just take the modulus. // e.g. [0x105, 0x107] & 0xff == [5, 7] // [0x105, 0x207] & 0xff == [0, 0xff] uinteger_t minChunk = imin.value & ~mask; uinteger_t maxChunk = imax.value & ~mask; if (minChunk == maxChunk && imin.negative == imax.negative) { imin.value &= mask; imax.value &= mask; } else { imin.value = 0; imax.value = mask; } imin.negative = imax.negative = false; return this; } IntRange castDchar() { // special case for dchar. Casting to dchar means "I'll ignore all // invalid characters." castUnsigned(0xFFFFFFFFUL); if (imin.value > 0x10FFFFUL) // ?? imin.value = 0x10FFFFUL; // ?? if (imax.value > 0x10FFFFUL) imax.value = 0x10FFFFUL; return this; } IntRange _cast(Type type) { if (!type.isintegral() || type.toBasetype().ty == Tvector) return this; else if (!type.isunsigned()) return castSigned(type.sizemask()); else if (type.toBasetype().ty == Tdchar) return castDchar(); else return castUnsigned(type.sizemask()); } IntRange castUnsigned(Type type) { if (!type.isintegral() || type.toBasetype().ty == Tvector) return castUnsigned(ulong.max); else if (type.toBasetype().ty == Tdchar) return castDchar(); else return castUnsigned(type.sizemask()); } bool contains(IntRange a) { return imin <= a.imin && imax >= a.imax; } bool containsZero() const { return (imin.negative && !imax.negative) || (!imin.negative && imin.value == 0); } IntRange absNeg() const { if (imax.negative) return this; else if (!imin.negative) return IntRange(-imax, -imin); else { SignExtendedNumber imaxAbsNeg = -imax; return IntRange(imaxAbsNeg < imin ? imaxAbsNeg : imin, SignExtendedNumber(0)); } } IntRange unionWith(const ref IntRange other) const { return IntRange(imin < other.imin ? imin : other.imin, imax > other.imax ? imax : other.imax); } void unionOrAssign(IntRange other, ref bool union_) { if (!union_ || imin > other.imin) imin = other.imin; if (!union_ || imax < other.imax) imax = other.imax; union_ = true; } ref const(IntRange) dump(const(char)* funcName, Expression e) const return { printf("[(%c)%#018llx, (%c)%#018llx] @ %s ::: %s\n", imin.negative?'-':'+', cast(ulong)imin.value, imax.negative?'-':'+', cast(ulong)imax.value, funcName, e.toChars()); return this; } void splitBySign(ref IntRange negRange, ref bool hasNegRange, ref IntRange nonNegRange, ref bool hasNonNegRange) const { hasNegRange = imin.negative; if (hasNegRange) { negRange.imin = imin; negRange.imax = imax.negative ? imax : SignExtendedNumber(-1, true); } hasNonNegRange = !imax.negative; if (hasNonNegRange) { nonNegRange.imin = imin.negative ? SignExtendedNumber(0) : imin; nonNegRange.imax = imax; } } IntRange opUnary(string op:"~")() const { return IntRange(~imax, ~imin); } IntRange opUnary(string op : "-")() { return IntRange(-imax, -imin); } // Credits to Timon Gehr for the algorithms for &, | // https://github.com/tgehr/d-compiler/blob/master/vrange.d IntRange opBinary(string op : "&")(IntRange rhs) const { // unsigned or identical sign bits if ((imin.negative ^ imax.negative) != 1 && (rhs.imin.negative ^ rhs.imax.negative) != 1) { return IntRange(minAnd(this, rhs), maxAnd(this, rhs)); } IntRange l = IntRange(this); IntRange r = IntRange(rhs); // both intervals span [-1,0] if ((imin.negative ^ imax.negative) == 1 && (rhs.imin.negative ^ rhs.imax.negative) == 1) { // cannot be larger than either l.max or r.max, set the other one to -1 SignExtendedNumber max = l.imax.value > r.imax.value ? l.imax : r.imax; // only negative numbers for minimum l.imax.value = -1; l.imax.negative = true; r.imax.value = -1; r.imax.negative = true; return IntRange(minAnd(l, r), max); } else { // only one interval spans [-1,0] if ((l.imin.negative ^ l.imax.negative) == 1) { swap(l, r); // r spans [-1,0] } auto minAndNeg = minAnd(l, IntRange(r.imin, SignExtendedNumber(-1))); auto minAndPos = minAnd(l, IntRange(SignExtendedNumber(0), r.imax)); auto maxAndNeg = maxAnd(l, IntRange(r.imin, SignExtendedNumber(-1))); auto maxAndPos = maxAnd(l, IntRange(SignExtendedNumber(0), r.imax)); auto min = minAndNeg < minAndPos ? minAndNeg : minAndPos; auto max = maxAndNeg > maxAndPos ? maxAndNeg : maxAndPos; auto range = IntRange(min, max); return range; } } // Credits to Timon Gehr for the algorithms for &, | // https://github.com/tgehr/d-compiler/blob/master/vrange.d IntRange opBinary(string op : "|")(IntRange rhs) const { // unsigned or identical sign bits: if ((imin.negative ^ imax.negative) == 0 && (rhs.imin.negative ^ rhs.imax.negative) == 0) { return IntRange(minOr(this, rhs), maxOr(this, rhs)); } IntRange l = IntRange(this); IntRange r = IntRange(rhs); // both intervals span [-1,0] if ((imin.negative ^ imax.negative) == 1 && (rhs.imin.negative ^ rhs.imax.negative) == 1) { // cannot be smaller than either l.min or r.min, set the other one to 0 SignExtendedNumber min = l.imin.value < r.imin.value ? l.imin : r.imin; // only negative numbers for minimum l.imin.value = 0; l.imin.negative = false; r.imin.value = 0; r.imin.negative = false; return IntRange(min, maxOr(l, r)); } else { // only one interval spans [-1,0] if ((imin.negative ^ imax.negative) == 1) { swap(l, r); // r spans [-1,0] } auto minOrNeg = minOr(l, IntRange(r.imin, SignExtendedNumber(-1))); auto minOrPos = minOr(l, IntRange(SignExtendedNumber(0), r.imax)); auto maxOrNeg = maxOr(l, IntRange(r.imin, SignExtendedNumber(-1))); auto maxOrPos = maxOr(l, IntRange(SignExtendedNumber(0), r.imax)); auto min = minOrNeg < minOrPos ? minOrNeg : minOrPos; auto max = maxOrNeg > maxOrPos ? maxOrNeg : maxOrPos; auto range = IntRange(min, max); return range; } } IntRange opBinary(string op : "^")(IntRange rhs) const { return this & ~rhs | ~this & rhs; } IntRange opBinary(string op : "+")(IntRange rhs) { return IntRange(imin + rhs.imin, imax + rhs.imax); } IntRange opBinary(string op : "-")(IntRange rhs) { return IntRange(imin - rhs.imax, imax - rhs.imin); } IntRange opBinary(string op : "*")(IntRange rhs) { // [a,b] * [c,d] = [min (ac, ad, bc, bd), max (ac, ad, bc, bd)] SignExtendedNumber[4] bdy; bdy[0] = imin * rhs.imin; bdy[1] = imin * rhs.imax; bdy[2] = imax * rhs.imin; bdy[3] = imax * rhs.imax; return IntRange.fromNumbers4(bdy.ptr); } IntRange opBinary(string op : "/")(IntRange rhs) { // Handle divide by 0 if (rhs.imax.value == 0 && rhs.imin.value == 0) return widest(); // Don't treat the whole range as divide by 0 if only one end of a range is 0. // Issue 15289 if (rhs.imax.value == 0) { rhs.imax.value--; } else if(rhs.imin.value == 0) { rhs.imin.value++; } if (!imin.negative && !imax.negative && !rhs.imin.negative && !rhs.imax.negative) { return IntRange(imin / rhs.imax, imax / rhs.imin); } else { // [a,b] / [c,d] = [min (a/c, a/d, b/c, b/d), max (a/c, a/d, b/c, b/d)] SignExtendedNumber[4] bdy; bdy[0] = imin / rhs.imin; bdy[1] = imin / rhs.imax; bdy[2] = imax / rhs.imin; bdy[3] = imax / rhs.imax; return IntRange.fromNumbers4(bdy.ptr); } } IntRange opBinary(string op : "%")(IntRange rhs) { IntRange irNum = this; IntRange irDen = rhs.absNeg(); /* due to the rules of D (C)'s % operator, we need to consider the cases separately in different range of signs. case 1. [500, 1700] % [7, 23] (numerator is always positive) = [0, 22] case 2. [-500, 1700] % [7, 23] (numerator can be negative) = [-22, 22] case 3. [-1700, -500] % [7, 23] (numerator is always negative) = [-22, 0] the number 22 is the maximum absolute value in the denomator's range. We don't care about divide by zero. */ irDen.imin = irDen.imin + SignExtendedNumber(1); irDen.imax = -irDen.imin; if (!irNum.imin.negative) { irNum.imin.value = 0; } else if (irNum.imin < irDen.imin) { irNum.imin = irDen.imin; } if (irNum.imax.negative) { irNum.imax.negative = false; irNum.imax.value = 0; } else if (irNum.imax > irDen.imax) { irNum.imax = irDen.imax; } return irNum; } IntRange opBinary(string op : "<<")(IntRange rhs) { if (rhs.imin.negative) { rhs = IntRange(SignExtendedNumber(0), SignExtendedNumber(64)); } SignExtendedNumber lower = imin << (imin.negative ? rhs.imax : rhs.imin); SignExtendedNumber upper = imax << (imax.negative ? rhs.imin : rhs.imax); return IntRange(lower, upper); } IntRange opBinary(string op : ">>")(IntRange rhs) { if (rhs.imin.negative) { rhs = IntRange(SignExtendedNumber(0), SignExtendedNumber(64)); } SignExtendedNumber lower = imin >> (imin.negative ? rhs.imin : rhs.imax); SignExtendedNumber upper = imax >> (imax.negative ? rhs.imax : rhs.imin); return IntRange(lower, upper); } IntRange opBinary(string op : ">>>")(IntRange rhs) { if (rhs.imin.negative) { rhs = IntRange(SignExtendedNumber(0), SignExtendedNumber(64)); } return IntRange(imin >> rhs.imax, imax >> rhs.imin); } IntRange opBinary(string op : "^^")(IntRange rhs) { // Not yet implemented assert(0); } private: // Credits to Timon Gehr maxOr, minOr, maxAnd, minAnd // https://github.com/tgehr/d-compiler/blob/master/vrange.d static SignExtendedNumber maxOr(const IntRange lhs, const IntRange rhs) { uinteger_t x = 0; auto sign = false; auto xor = lhs.imax.value ^ rhs.imax.value; auto and = lhs.imax.value & rhs.imax.value; auto lhsc = IntRange(lhs); auto rhsc = IntRange(rhs); // Sign bit not part of the .value so we need an extra iteration if (lhsc.imax.negative ^ rhsc.imax.negative) { sign = true; if (lhsc.imax.negative) { if (!lhsc.imin.negative) { lhsc.imin.value = 0; } if (!rhsc.imin.negative) { rhsc.imin.value = 0; } } } else if (lhsc.imin.negative & rhsc.imin.negative) { sign = true; } else if (lhsc.imax.negative & rhsc.imax.negative) { return SignExtendedNumber(-1, false); } for (uinteger_t d = 1LU << (8 * uinteger_t.sizeof - 1); d; d >>= 1) { if (xor & d) { x |= d; if (lhsc.imax.value & d) { if (~lhsc.imin.value & d) { lhsc.imin.value = 0; } } else { if (~rhsc.imin.value & d) { rhsc.imin.value = 0; } } } else if (lhsc.imin.value & rhsc.imin.value & d) { x |= d; } else if (and & d) { x |= (d << 1) - 1; break; } } auto range = SignExtendedNumber(x, sign); return range; } // Credits to Timon Gehr maxOr, minOr, maxAnd, minAnd // https://github.com/tgehr/d-compiler/blob/master/vrange.d static SignExtendedNumber minOr(const IntRange lhs, const IntRange rhs) { return ~maxAnd(~lhs, ~rhs); } // Credits to Timon Gehr maxOr, minOr, maxAnd, minAnd // https://github.com/tgehr/d-compiler/blob/master/vrange.d static SignExtendedNumber maxAnd(const IntRange lhs, const IntRange rhs) { uinteger_t x = 0; bool sign = false; auto lhsc = IntRange(lhs); auto rhsc = IntRange(rhs); if (lhsc.imax.negative & rhsc.imax.negative) { sign = true; } for (uinteger_t d = 1LU << (8 * uinteger_t.sizeof - 1); d; d >>= 1) { if (lhsc.imax.value & rhsc.imax.value & d) { x |= d; if (~lhsc.imin.value & d) { lhsc.imin.value = 0; } if (~rhsc.imin.value & d) { rhsc.imin.value = 0; } } else if (~lhsc.imin.value & d && lhsc.imax.value & d) { lhsc.imax.value |= d - 1; } else if (~rhsc.imin.value & d && rhsc.imax.value & d) { rhsc.imax.value |= d - 1; } } auto range = SignExtendedNumber(x, sign); return range; } // Credits to Timon Gehr maxOr, minOr, maxAnd, minAnd // https://github.com/tgehr/d-compiler/blob/master/vrange.d static SignExtendedNumber minAnd(const IntRange lhs, const IntRange rhs) { return ~maxOr(~lhs, ~rhs); } static swap(ref IntRange a, ref IntRange b) { auto aux = a; a = b; b = aux; } }