Hi John,
I think your change is correct. The loop can also be optimized a little if
we check beforehand for the completely whitespace case, so I'm checking in
the patch below.
I noticed you cc'd the libgcj patches list: this problem does not apply to
libgcj, because it has a different (native) implementation of String.
regards
[ bryce ]
"[EMAIL PROTECTED]" wrote:
> I had problems with the String.trim() method, changing it from:
[...]
> Note I just added "+ 1" in the return line.
>
> I haven't tested this thoroughly but it made my problems go away.
2000-05-15 Bryce McKinlay <[EMAIL PROTECTED]>
* java/lang/String.java (trim): Optimize `end' loop. Don't lose the
last character of the returned string.
===================================================================
RCS file: /cvs/classpath/java/lang/String.java,v
retrieving revision 1.27
diff -u -r1.27 String.java
--- String.java 2000/05/09 02:12:26 1.27
+++ String.java 2000/05/15 09:10:29
@@ -905,14 +905,18 @@
if (count == 0 || (value[0] > '\u0020' && value[count-1] > '\u0020'))
return this;
int begin = 0;
- for (; begin < count; begin++)
- if (value[begin] > '\u0020')
- break;
- int end = count-1;
- for (; end >= 0; end--)
- if (value[end] > '\u0020')
- break;
- return substring(begin, end);
+ for (;; begin++)
+ {
+ if (begin == count)
+ return "";
+ if (value[begin] > '\u0020')
+ break;
+ }
+ int end = count;
+ for (;;)
+ if (value[--end] > '\u0020')
+ break;
+ return substring(begin, end + 1);
}
/**
