I briefly glanced at this today and will look into it more tonight, I noticed that we did not do a dmr release upgrade for as yet, so for sure it's not the base64 routine. This also means the flush issue is still there.

Sent from my iPhone

On Mar 21, 2011, at 5:20 PM, Jonathan Pearlin <jpearlin1@gmail.com> wrote:

From the testing I was able to do with the string provided (and some that I generated), it appears that everything works fine if the Base64 encoded string is complete.  The string provided definitely appears to have been cut off.  I feel as though we are looking in the wrong place:  we should try to see if we can re-create whatever caused the response to be cut short (is it possible that the HTML response was cut off, which caused the whole problem or can you repeat this every time)?  My suspicion is that we will find something wrong with the encoding side.

--Jonathan

On Mon, Mar 21, 2011 at 11:49 AM, Heiko Braun <hbraun@redhat.com> wrote:

Yeah, that's what I see as well.
But I cannot identify the changes that could have caused this problem.

Ike

On Mar 19, 2011, at 3:57 AM, Jason Greene wrote:

>> Second, it appears that the Base64 encoded string is cut off