Remy Maucherat wrote:
Jean-Francois Arcand wrote:
Can you build a test case for the issue ? That would be easier than trying to guess what's going on (and it must be an obvious bug if there's a bug, like for the input stream reading from the char buffer ...). You've got to be able to reproduce it without using JSPs or tags if it's a valid bug, right ?
-1 about the NPE change. I don't want to add null checks in every method to handle the one broken webapp. This doesn't make any sense.
OTOH, here's something I would accept if you want to do that cleanup: the clear() would associate the facade with a (static final) HTTPServletRequest implementation which would return an ISE for every method.
And do the same for the response facade.
I think it's not worth the pain, so I would be +0 for the change.
I realized some time later that this doesn't work, as the response facade uses methods from CoyoteResponse (for the response commit state).
Since I can't provide an acceptable solution, I'll have to accept your polish if you think it is useful, and I change my vote to 0.
Remy
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
