• via Digitalmars-d-bugs (70/70) Nov 28 2016 https://issues.dlang.org/show_bug.cgi?id=16800

via Digitalmars-d-bugs <digitalmars-d-bugs puremagic.com> writes:

https://issues.dlang.org/show_bug.cgi?id=16800

          Issue ID: 16800
           Summary: Minimum double inconsistency/problems + double.min +
                    (number '4.9E-324' is not representable) / Mac /
                    Windows
           Product: D
           Version: D2
          Hardware: x86
                OS: Mac OS X
            Status: NEW
          Severity: enhancement
          Priority: P1
         Component: dmd
          Assignee: nobody puremagic.com
          Reporter: soywiz gmail.com

Double literal 4.9E-324 produces error in dmd/ldc2 for mac, while it compiles
on windows or even on wine without any issue.

Also double.min seems to not be defined, while double.max is.
https://dlang.org/ctod.html double.min appears in this page, so in the case it
is not defined it should be removed as example in that page.

D doubles are IEEE 754 right? https://dlang.org/d-floating-point.html
At least literals should work consistently among platforms?

Since I'm generating double literals automatically I need a reliable way to
represent doubles that compile on every platform D supports. So what should I
do?
Something like this?
if (doubleValue >= 0.0 && doubleValue <= 4.940656e-324) "4.940656e-324" else
"$doubleValue"
since double.min is not defined I can't do this:
if (doubleValue >= 0.0 && doubleValue <= 4.940656e-324) "double.min" else
"$doubleValue"


DMD version (in both cases): v2.072.0

ldc2 --version
LDC - the LLVM D compiler (3461e0):
  based on DMD v2.070.2 and LLVM 3.9.0

Fails on MacOSX:

➜  jtransc-d git:(master) ✗ dmd test.d
test.d(15): Error: number '4.9E-324' is not representable
➜  jtransc-d git:(master) ✗ ldc2 test.d
test.d(15): Error: number '4.9E-324' is not representable

Works on Windows + wine:

➜  jtransc-d git:(master) ✗ wine cmd
dmd testMicrosoft Windows 5.1.2600 (1.8.5)

Z:\Users\soywiz\Projects\jtransc-examples\hello-world\build\jtransc-d>dmd
test.d

Z:\Users\soywiz\Projects\jtransc-examples\hello-world\build\jtransc-d>exit

// Tested code:
import std.stdio;

int main() {
//      long v = 1L;
        double a = double.max;
//      double a = double.min;
        writefln("%e", a);
/*
http://docs.oracle.com/javase/8/docs/api/java/lang/Double.html#MIN_VALUE
MIN_VALUE
public static final double MIN_VALUE
A constant holding the smallest positive nonzero value of type double, 2-1074.
It is equal to the hexadecimal floating-point literal 0x0.0000000000001P-1022
and also equa$
*/
        long v = 1L;
        double b = 4.9E-324;
        //align(8) long v = 0x7ff8000000000000L;
        //long v = 0x0010000000000000L;
//      double b = *cast(double*)&v;
        writefln("%e", b);
        return 0;
}

--