> Here's the header information (from wget -s http://totale.usc.edu):
Wget doesn't give you the right info; it's following a redirect to https://totale.usc.edu/, which speaks correct protocol. Using either telnet or curl -i, you can get at the original data: $ telnet totale.usc.edu 80 Trying 128.125.253.94... Connected to totale.usc.edu. Escape character is '^]'. GET / HTTP/1.1 Host: totale.usc.edu HTTP/1.1 302 Object Moved Server: NS_3.0 Location: https://totale.usc.edu/ Content Type: text/html Cache Control: private Connection: close ... As you can see, both the Content-Type and Cache-Control headers are lacking the dash. According to the grammar in RFC 2616 Sections 4.2 and 2.2, a field name cannot contain a space. This is a bug in the server (NS_3.0, whatever that is), not in Polipo. > On http://linux.bkbits.net:8080/linux-2.5, I get: As mentioned before, this web server is completely broken. For a GET request, it serves a raw date in the headers area. For a HEAD request, it gives an error reply that doesn't even vaguely resemble anything like HTTP. Tom, could you please downgrade this bug report to ``wishlist''? Greg, any chance you could yell at the people at totale.usc.edu to stop using broken software? (Falk, there's no point yelling at the people at BitMovers, as far as I can tell.) > What's the latest on this issue? I get the feeling it results in > real loss of functionality for some of my users. Users are good, I agree. As a workaround, the head branch ignores header lines it cannot parse (by default; set laxHttpParser to false to return to the previous behaviour). This is not a good solution, as that might get the HTTP parser out of sync and therefore cause parsing errors further down the line, but it would more or less work around both issues described above. As I'm a little late getting 0.10 out the door, I'll backport the workaround to the 0.9 branch as soon as I've got a chance. For the time being, all I can suggest is that Falk and Greg might want to switch to the head branch (which should be fairly stable as long as you don't use a redirector). Sorry for that. Juliusz -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
