Hi,
inspired by Laurens' experiment with Accept I just did some tests with
Content-Type.
When setting Content-Type twice, IE7 behaves as specified (sending a
*broken* Content-Type header), while Firefox3 simply sends the second
value that was set.
So apparently, the two most widely used browsers differ in what they do.
Is it really a good idea to standardize on what IE does, *knowing* that
it may cause broken requests to be generated?
BR, Julian
PS: it's a bit ironic if the same group of people standardizes APIs that
make it dead easy to generate invalid headers then complains to the HTTP
WG for the lack of defined behaviour for broken requests...