Re: bug 10491 was Re: More astonishing progress in japi scores

[Robert Schuster]

Hi
the mailer is funny. I got Sven's response to Andrew's mail but not Andrew's.

Sven de Marothy wrote:

Andrew Haley writes:
  

Has anyone seen that one: 
http://savannah.gnu.org/bugs/?func=detailitem&item_id=10491
I recently discovered it while writing some test files for the 
XMLDecoder I am working on. In an attempt to find the reason I
      

followed 
  

to the C source that does the conversion (java_lang_Double.c) and its
obvious to me that the code does not care about the special values: 
[+-]NaN, [+-]Infinity
      

What is really wrong here?  The bug has no test case, and doesn't
really explain.
    

I think he's talking about Double.parseDouble(), an input string of
"NaN" should return a NaN double. Now, it throws an exception.

One-liner: 
System.out.println(Double.parseDouble("NaN"));
  

Sorry, it was late and I was a bit unattentive when writing the bug report. I added the base for test cases showing the
defect and what was expected instead now.

Although, personally.. I'm not so sure of re-writing it in Java. Seems
like overkill for a relatively trivial bug. 
(Of course, you could have both too, and add a build-time option. )

Having both as a compile time option seems to be in most people's interest (IMHO). I could fix the problem in Java and C but dont know how to make it
a compile time option. Help?

cu
Robert