I am using a mod_perl content handler to implement a series of forms. The first request from a particular browser is always a "GET" that provides some required parameters in the URL. From then on, all parameters are passed as hidden form variables, and only POSTs should occur.
If the browser accepts cookies, a unique session ID is stored in a long lived cookie and is also logged in the access log, strictly for tracking (no session data associated with it). This is not 100% reliable because not everyone accepts cookies, but it works enough of the time that I can track most of the requests by this ID. 99.9% (roughly) of the time, everything works fine and the correct sequence is followed: GET /formseries?a=123&b=234 POST /formseries (a=123, b=234 passed in post vars) POST /formseries (a=123, b=234 passed in post vars) POST /formseries (a=123, b=234 passed in post vars) POST /formseries (a=123, b=234 passed in post vars) POST /formseries (a=123, b=234 passed in post vars) The problem is that there are a few request trails per day, where the following sequence is followed instead: GET /formseries?a=123&b=234 POST /formseries (a=123, b=234 passed in post vars) POST /formseries (a=123, b=234 passed in post vars) POST /formseries (a=123, b=234 passed in post vars) GET /formseries (required parameters not passed on this request) The GET fails because the handler does not have the required parameters it needs to pull info from the database. An incomplete of user agents that the problem occurs with are: Mozilla/4.0 (compatible; MSIE 5.0; Windows 98; DigExt) Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0) Mozilla/4.0 (compatible; MSIE 5.5; Windows 98; Win 9x 4.90) One other piece of evidence is that the "Referer" is empty on the erroneous GET request. I have tried but been unable to reproduce the failing request sequence using IE, even using back, forward, and reload button in a number of sequences. My questions are, is there some IE anomaly or some anomalous way of configuring or using IE that can cause it to all of a sudden do a GET instead of a POST (anything short of manually entering the url)? Could a proxy be causing the problem (a small subset of the failing requests to seem to be coming through proxies)? Also, is there anything more I could track that would give more insight into the problem? Currently the content handler does detect that the required parms are missing and then dumps the complete request and timestamp into the error log. Also, please let me know if there is a more appropriate forum to ask this question. -- Matthew Pressly
