[Mono-bugs] [Bug 467201] [PATCH] Improper conv.r4, conv.r8 and float return implementations

[bugzilla_noreply at novell.com]

https://bugzilla.novell.com/show_bug.cgi?id=467201

User kornelpal at gmail.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=467201#c10


Kornél Pál <kornelpal at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #10 from Kornél Pál <kornelpal at gmail.com>  2009-02-20 17:03:54 MST ---
The thing is that it's actually 80-bits.

64-bit rounding is only applied to arithmetic operations. When you load
integers from the stack they are converted to full 80-bits.

Last year I've contacted the ECMA technical group TG3 suggesting that conv.r.un
should be renamed to conv.r8.un because the name conv.r.un suggests that it's
using the native float size but the current specs state that it converts to
either float32 or float64 and their position is that this is really intended to
convert integers to the native float size.

Nevertheles MS.NET implements it as if it were conv.r8.un.

But your comment made me realize that I missed an important test case.

-- 
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.