This is definitely getting more and more ridiculous. I am somewhat of an MS basher myself but I was amazed to see the number of people who claim to have an expert opinion without even testing something as simple as this. I tested two browsers - IE 6.0 and Netscape 4.77 against a Netscape Web server running (tried to get a simple HTML page) on Sol 8. I observed both the browsers initiate the connection as specified in the RFC. (i.e.) a Syn was sent each time by the client. One anomaly which I could not explain was the way both the browsers treated the connection termination.
After the server responded with the HTTP 200 for a GET request, the server sends out a Fin Ack for which the client (Netscape) responds with an Ack and the server subsequently moved to a Fin_Wait_2 state. This seems to hang in there infinitely until another connection is opened up from the client machine. In this case the client has not sent a Fin.
In the case of IE, the servers' Fin Ack was answered with an Ack and a Rst. It is interesting to note that the server connection is promptly closed and there are no hanging connections. This definitely goes against what the post originated as and more. This could be improper handling but this certainly gets the job done.
My question (definitely attributed to my ignorance) is... what effect does the client OS have to play in this ? Does the browser come entirely with its own TCP implementation or does it utilize the OS TCP implementation ? (If this question in itself is stupid.... any RTFM pointers appreciated)
Slashdot Mirror
This is definitely getting more and more ridiculous. I am somewhat of an MS basher myself but I was amazed to see the number of people who claim to have an expert opinion without even testing something as simple as this. I tested two browsers - IE 6.0 and Netscape 4.77 against a Netscape Web server running (tried to get a simple HTML page) on Sol 8. I observed both the browsers initiate the connection as specified in the RFC. (i.e.) a Syn was sent each time by the client. One anomaly which I could not explain was the way both the browsers treated the connection termination.
... what effect does the client OS have to play in this ? Does the browser come entirely with its own TCP implementation or does it utilize the OS TCP implementation ? (If this question in itself is stupid .... any RTFM pointers appreciated)
After the server responded with the HTTP 200 for a GET request, the server sends out a Fin Ack for which the client (Netscape) responds with an Ack and the server subsequently moved to a Fin_Wait_2 state. This seems to hang in there infinitely until another connection is opened up from the client machine. In this case the client has not sent a Fin.
In the case of IE, the servers' Fin Ack was answered with an Ack and a Rst. It is interesting to note that the server connection is promptly closed and there are no hanging connections. This definitely goes against what the post originated as and more. This could be improper handling but this certainly gets the job done.
My question (definitely attributed to my ignorance) is