Re: PHP 5.2 + IE 7 = HTTP 304 in login procedure [SOLVED]

The problem wasn't quite there. The 304 response was correct and didn't really cause a problem in IE7. In fact, the problem was somewhere else. As mentioned by someone in the PHP doc comments (http://be.php.net/manual/en/ref.session.php#64125 ), IE is the only one to reject urls of the likes of http://example_demo.littleweb.com/ because of the "_" character. This means that basically IE will display the site without a problem, but will not register any cookie, and of course will not mention it, so it makes it all much easier to detect. The fact that FF and Opera get a session each does not help at all in finding the problem. Changing the URL to http://exampledemo.littleweb.com/ will work perfectly. Yannick Le dimanche 11 mars 2007 à 23:01 +0000, Yannick Warnier a écrit : > Hello, > > One of my clients is currently having a problem when logging into one of > my site. > Investigating further (because it works with Firefox with his login/pass > from my machine), it appears the problem is caused for an obscure reason > when IE7 requests the page and obviously does a "conditional GET", which > only loads what's necessary for this login to proceed when the page has > been updated since last time. The returned HTTP header is 304: Not > modified, which is not returned with other browsers (others get a 200 > header). > This is true for PHP pages as well as included CSS files, which triggers > the question of having any link to PHP at all... > > I've looked on Google quite a bit, and if I have found people having the > same kind of problems, they generally report it along with the fact that > they use incorrectly the header('HTTP/1.1 ...'); or > header('Status: ...'); function, so the fix is generally a change of > these. > > However, my application doesn't set any of these headers from inside the > PHP code. > > Before I start getting into the whole Apache2 config (which I'm not to > good at) and try a lot of funny things in a bid to discover one element > that would cause this, I'd like to know... > > Does anybody know the problem and have already found a fix? > > Thanks, > > Yannick >