On Jun 25, 2004, at 9:36 AM, Nick Kew wrote:
>
> OK, on further consideration after posting, I've reached the view that
> the 2.0.49 patch should be modified. I'd propose to run the same
> transformations on Set-Cookie headers from the backend, but move
> the rewriting - along with the rewriting of Date and Location headers
> (ap_proxy_date_canon and ap_proxy_location_reverse_map) into
> ap_proxy_read_headers. That actually both cleans up the code (reduces
> adhockery) and makes it more efficient (reduces table operations),
> as well as making my patch orthogonal to yours.
>
The patch was never intended to be the full and complete
solution, rather a specific fix to a specific problem. The
emails noted that ideally we could select which headers to
honor/keep in a proxied environment instead of hardcoding
the assumption of "just" cookies.
See #27023
|