Re: [Java] Call "fold" in convert_ieee_real_to_integer

[Ranjit Mathew <rmathew at gmail dot com>]

On Wed, 21 Jul 2004 15:23:06 -0600 (MDT), Roger Sayle
<roger@eyesopen.com> wrote:
> 
> On Wed, 21 Jul 2004, Ranjit Mathew wrote:
> > No, "case (int)Float.NaN:" also doesn't work.
> 
> Hi Ranjit,
> 
> Could you check whether the patch below cures the recent JACKS
> failures (and "case (int)Float.Nan") for you?
> 
> This following patch has been tested on i686-pc-linux-gnu, with a full
> "make bootstrap", all default languages including java, and regression
> tested with a top-level "make -k check" with no new failures.
> 
> Ok for mainline if this fixes Ranjit's problems?

Thanks a tonne for looking into this Roger!

Unfortunately, there still is some problem and
quite curiously, the behaviour is different for "-C" 
(generate Java bytecode) and "-fsyntax-only"
and "-c" (generate object code) as can be
seen from the following:
---------------- 8< ----------------
~/tmp > cat X.java
public class X
{
  public void foo( int i)
  {
    switch(i)
    {
    case ((int )Float.NaN == 0 ? 1 : 0): break;
    case (int )Float.NaN: break;
    }
  }
}
~/tmp > $MYGCJ -fsyntax-only X.java
~/tmp > $MYGCJ -c X.java
~/tmp > $MYGCJ -C X.java
X.java: In class `X':
X.java: In method `X.foo(int)':
X.java:8: error: duplicate case label: `0'.
       case (int )Float.NaN: break;
       ^
X.java:7: error: original label is here.
       case ((int )Float.NaN == 0 ? 1 : 0): break;
       ^
2 errors
~/tmp >
---------------- 8< ----------------

Thanks,
Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT gmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/