To my surprise, I just discovered that ResponseInputStream is already
attempting to support chunked transfer encoding. Therefore, the patch I
previously submitted to address this bug is moot. Please disregard my
previous patch postings for bug 5018.
However, I have identified the bug in ResponseInputStream. At line 99,
it attempts to find the index of the constant "chunked" in a
case-sensitive manner. Some servers, such as WebLogic, reply with
"Chunked", thus causing this test to fail. The attached patch should
solve this problem.
My apologies to those who have invested any time in reviewing my
previous patch. Chalk it up to someone who is not initimately familiar
with the code (yet).
Yours truly,
Paul C. Bryan <[EMAIL PROTECTED]>
http://pbryan.net/
--- ResponseInputStream.java.orig Sat Dec 1 21:23:09 2001
+++ ResponseInputStream.java Sat Dec 1 21:23:29 2001
@@ -96,7 +96,7 @@
// Retrieving transfer encoding header
Header transferEncoding = method.getResponseHeader("transfer-encoding");
if((null != transferEncoding)
- && (transferEncoding.getValue().indexOf("chunked") != -1))
+ && (transferEncoding.getValue().toLowerCase().indexOf("chunked") != -1))
chunk = true;
// Retrieving content length header
--
To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
