Justin Erenkrantz wrote:
As I just noted in STATUS for 2.0, read_length isn't a sufficient check. It'd only be set if the client has *already* read the body *and* they used the 1.3.x mechanisms for reading the request body.
both true in 100% of the cases I've seen in the wild. ok, I'll admit it is only one case :)
The proper check is to look for Transfer-Encoding and Content-Length in the request headers. This is what ap_http_filter in http_filters.c does to look for a request body. -- justin
I'm a little concerned about path length, but I guess subrequests aren't as critical. unfortunately a lot of headers start with C so the Content-Length table hash isn't as quick as it could be.
Greg
