I just ran into another site where this goes wrong (http://www.canon.co.jp/Imaging/PSG1/PSG1_Firmware-e.html) with the latest & greatest 2.7b (plain, no additional patches like in the version I initially talked about).
I didn't find any followups to this in my mailbox, so I wonder what the status is? Here's the original message as a reminder; in addition I've added some new testing results at the end. On Sat 02 Mar 2002, Paul Slootman wrote: > I've noticed sometimes that when I request a page that's already in the > cache because I accessed it in another browser, it won't load properly. > This is when a page won't work in Opera (my main browser) because of > javascript problems, and then I try it in Mozilla or Netscape. Mozilla > shows a blank page (show source shows <html><body></body></html>, but > I've since learnt here that Mozilla does that to empty pages, grrr). > Netscape says "communications error". > > For no particular reason I tried telnetting to wwwoffle to try it by > hand, after wget didn;t have any problems getting the page through > wwwoffle. I saw this: > > $ telnet localhost 8080 > Trying 127.0.0.1... > Connected to localhost. > Escape character is '^]'. > GET http://www.digadoka.com/index3.htm HTTP/1.0 > > HTTP/1.0 200 OK > Date: Sat, 02 Mar 2002 21:24:18 GMT > Server: Apache/1.3.20 (Unix) PHP/4.0.5 > Last-Modified: Thu, 14 Feb 2002 01:38:43 GMT > ETag: "1cd339-532-3c6b1523" > Accept-Ranges: bytes > Content-Type: text/html > X-Pad: avoid browser bug > Content-Encoding: gzip > Content-Length: 1333 > Connection: close > Proxy-Connection: close > > <html> > > <head> > <META HTTP-EQUIV="Content-type" CONTENT="text/html; charset=ISO-8859-1"> > .... > > > > Note how wwwoffle says "Content-Encoding: gzip" even though I didn't ask > for it, but then proceeds to output plain text. This is probably what's > confusing the hell out of Mozilla and Netscape. If I delete the > www.digadoka.com cache directory, it works fine in netscape. > > I should mention that this version of wwwoffle contains the two patches > that went by here recently (the not-modified and the > request-compressed-data patches). I haven't yet tried backing out these > patches to see if that changes things, I thought I should tell this > first. I'll try to look at it Sunday. As I said, the last paragraph isn't applicable any more, as this is vanilla wwwoffle 2.7b. Here's what I now get: $ nc localhost 8080 > /tmp/ps1 get http://www.canon.co.jp/Imaging/PSG1/PSG1_Firmware-e.html HTTP/1.0 $ less /tmp/ps1 HTTP/1.0 200 OK Date: Wed, 24 Apr 2002 16:48:39 GMT Server: Apache/1.3.14 (Unix) Content-Type: text/html Content-Encoding: gzip Content-Length: 53407 Connection: close Proxy-Connection: close <html>^M^M <head>^M <meta http-equiv="content-type" content= [...] Interesting detail: $ nc www.canon.co.jp 80 > /tmp/ps1 GET /Imaging/PSG1/PSG1_Firmware-e.html HTTP/1.0 Host: www.canon.co.jp Accept-Encoding: gzip, deflate, compress;q=0.9 $ less /tmp/ps1 HTTP/1.1 200 OK Date: Wed, 24 Apr 2002 17:03:32 GMT Server: Apache/1.3.4 (Unix) Transfer-Encoding: chunked Content-Type: text/html 1000 <html>^M^M <head>^M <meta http-equiv="content-type" content= [...] Note how I request compressed data, but don't get it (I get "chunked" instead?!). So why wwwoffle add the Content-Encoding: gzip header? Paul Slootman
