[issue10751] REMOTE_USER and Remote-User collision in wsgiref

2011-01-04 Thread Alex Raitz
Alex Raitz added the comment: I had previously tested it against simple_server. However, in reviewing my test, I realized that you are correct that wsgiref headers is not misbehaving. It appears that in simple_server, the values of remote-user and remote_user both end up in

[issue10751] REMOTE_USER and Remote-User collision in wsgiref

2011-01-04 Thread Alex Raitz
Alex Raitz added the comment: Per the first line of my previous comment, please ignore HTTP_REMOTE_USER. The risk is that if the proxy does not place the user-supplied 'remote-user=VALUE1' before the proxy-supplied 'REMOTE_USER=VALUE2', wsgiref will overload REMOTE_US

[issue10751] REMOTE_USER and Remote-User collision in wsgiref

2011-01-04 Thread Alex Raitz
Alex Raitz added the comment: Yes, I was referring to REMOTE_USER, apologies for the conflation with HTTP_REMOTE_USER, which was one of the HTTP headers that a proxy which we were testing was setting. The customer that reported this issue to us was using FireFox with Tamper Data to set

[issue10751] WSGIREF - REMOTE_USER and REMOTE-USER collision

2010-12-21 Thread Alex Raitz
New submission from Alex Raitz : Clients can overwrite 'REMOTE_USER' header variable value with an arbitrary 'Remote-User' value by specifying the later after the former. This has tricky implications when a proxy server is being used, namely that if the proxy passes a re