← Back to index | Original Bugzilla link

Bug 12831 – assert incorrectly fails on to!float conversion

Status: RESOLVED
Resolution: INVALID
Severity: normal
Priority: P1
Component: dmd
Product: D
Version: D2
Platform: x86
OS: All
Creation time: 2014-06-01T01:57:00Z
Last change time: 2014-06-01T13:23:38Z
Assigned to: nobody
Creator: mk

Comments

Comment #0 by mk — 2014-06-01T01:57:22Z

import std.conv; void main() { float a = 1.234f; float b = to!float("1.234"); assert (a == b); assert (a == to!float("1.234")); //fails } DMD32 D Compiler v2.066-devel-38c5c36 Works in gdc. Also works with real.

Comment #1 by mk — 2014-06-01T02:12:12Z

import std.conv; void main() { float a = 1.234f; real b = to!float("1.234"); assert (a == b); //fails } Maybe I'm missing something about fpu arithmetics, this fails too, but again works in gdc.

Comment #2 by mk — 2014-06-01T13:23:38Z

marking INVALID as floating point numbers shouldn't be compared directly for equality (machine and platform dependent).