Odi, NPE in HttpMethodBase also looks like a undisputable bug in HttpClient. Testing objects for being non-null is pretty much a must, as far as I can tell.
Tim, This said, to ease the acceptance of a patch, no matter how obvious the bug might be, a bug report and a suggested fix should be accompanied by a test case (JUnit in case of HttpClient) that fails without a fix and competes successfully with. Oleg -----Original Message----- From: Ortwin Glück [mailto:[EMAIL PROTECTED] Sent: Thursday, October 16, 2003 08:28 To: Commons HttpClient Project Subject: Re: [PATCH] Single line feed treated as false connection close Tim, thanks for tracking that down. I think the fix for the HttpParser is okay. The two other issues need proper review and testing. Unfortunately it's too late now (we are in RC2) to make API changes to HttpMethod. Please if you have the time, could you split the patch into three parts: 1. Parser fix 2. NPE in HttpMethodBase 3. enhancements Part 1 is trivial. But part 2 and 3 should be acompanied by a test case. And could you explain the goal of part 3, please? This patch will not make it into 2.0 but you are welcome to propose a change for CVS HEAD. Thanks Odi Tim McCune wrote: > This patch is against 2.0rc2. > I ran into a problem where the HttpClient kept thinking that the > connection was closed while talking to an IIS 5.0 server, even though my > web browser could load the page fine. Turns out that the server was > sending a line feed as the first byte of the response, which HttpClient > treats as a closed connection due to a bug in HttpParser. This patch > also contains a fix in HttpMethodBase.java for a NullPointerException > that happens when responseConnection is null, as well as an enhancement > that someone else in our organization submitted a while back but > apparently never got accepted. --------------------------------------------------------------------- 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]
