Fixes for Mauve JDK 1.0 tests

[Dalibor Topic]

Hi,

here is the last patch in the mauve JDK 1.0
compatibility series. This
one fixes the remaining[1] mauve bugs in
java.lang.Math, java.lang.String,
java.lang.StringBuffer.


* Math
there is a problem with the pow method, on i386-linux
using glibc
2.2.4 at least. Its output differs from what the spec
says for
 +-1.0 ** Inf.

I'm quite sure that this patch makes pow slower, since
the parameter
check is performed in java code. If that's a problem,
I'll try to
rewrite it in C.

* String
indexOf didn't throw NullPointerException if the
pattern String was
null. Now it does.

* StringBuffer
insert didn't allow Strings to be inserted after the
last
character. That's appending, basically. I've factored
out the index
checking code into two small private methods, and used
them to
eliminate duplicate index checking code in the class.

Performance effect: could be negative if the compiler
doesn't inline
private methods.

have fun

dalibor topic

[1] together with my double/float printing patch. You
might experience a failure in the Float test as well
if you use jikes, but that's due to a bug in jikes.


__________________________________________________
Do You Yahoo!?
Yahoo! Tax Center - online filing with TurboTax
http://taxes.yahoo.com/
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mauve.patch.changelog
Url: http://kaffe.org/pipermail/kaffe/attachments/20020404/65c57513/attachment-0003.txt 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mauve.patch
Type: text/x-diff
Size: 3257 bytes
Desc: mauve.patch
Url : http://kaffe.org/pipermail/kaffe/attachments/20020404/65c57513/attachment-0003.patch