ok, here is what I found,
CoyoteOutputStream.java
public void print(String s)
throws IOException {
ob.write(s);
}
This calls OutputBuffer.java
public void write(String s)
throws IOException {
if (suspended)
return;
if (s == null)
s = "null";
conv.convert(s);
conv.flushBuffer();
}
and hence the NPE
Filip
Filip Hanik - Dev Lists wrote:
I figured :). I'll revert and hunt down the root cause. My guess is
that if the response was used with output stream first, then recycled
and then used again, somehow the conv ends up being null and causes a
NPE in the write method.
Filip
Remy Maucherat wrote:
[EMAIL PROTECTED] wrote:
Author: fhanik
Date: Mon Nov 20 08:47:03 2006
New Revision: 477251
URL: http://svn.apache.org/viewvc?view=rev&rev=477251
Log:
TCK correction, depending on the sequence of the tests, the
converter turns out to be null at certain times.
Added in a check to create the converter even when getting the
output stream.
I have to veto this. It's using the output stream in that case, so
why does it matter if the converter (= the character encoder) is null ?
Rémy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
