From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.1+) Gecko/20010623 BuildID: 2001062306 If you go to the referenced URL, you get a properly displayed page of accesses to some software I publish. If you then visit any other page and go back, you get seemingly random corruption at the bottom of the page. Reproducible: Always Steps to Reproduce: 1.Visit http://www.arlut.utexas.edu/cgi-bin/log_display.pl?log_name=access&lines=500&grep=gash2 2. Go anywhere else 3. Hit back, then scroll down to the bottom of the page Actual Results: Garbage at the bottom of the page Expected Results: Redisplay the log page in identical fashion as when it was first viewed
Additional experimentation shows that viewing the page source for this page shows the corruption before ever leaving and returning to the page. Whatever source the 'view source' function is drawing on is corrupted.
Okay, yet more investigation shows that the page my CGI is generating is malformed, with no <HTML><BODY> at the top, but including </BODY></HTML> at the bottom. My bad, and I can fix this, but I'm going to leave it as is for the sake of shaking out Mozilla's behavior on this. If you'd rather I just go away, I'd understand and will fix my stuff and be happy. It does look like Mozilla is doing something unexpected here, though.
It works for me w2k 2001062204 and on linux 2001062406. The parser should be able to handle missing <html> and <body> tags, so if you are still seeing this with todays build see if you can make a testcase, or at least leave one page broken on your server. Also if you could describe what sort of corruption is happening that would help. Are the </body> and </html> tags appearing in the view source? What happens if you do a save as (view source goes through a slightly different path. Are you capable of looking at the HTTP headers? If things are showing up wrong on an initial load (beware of cache) the problem is probably in networking HTTP.
If the page length gets shorter, you might be seeing bug 79983.
I do see this with the current build (2001062306) on Linux. The page length gets longer, with a bunch of random garbage extracted from the contents of the page tacked onto the end. Doing 'save as' results in a file saved to disk with the corruption. I will upload it as an attachment to this bug.
Created attachment 39876 [details] HTML 'save as' showing the corruption that I see when viewing and hitting 'back' on the primary URL
I still see this behavior in today's (2001062406) Linux build, as well.
I looked at bug 79983, but it looks different enough that I don't recognize it as being the same bug. The fact that I can view the CGI output initially okay, but then see the corruption when I do a 'save as' or a 'view source', or going to another page and then hitting 'back' makes me think it's happening someplace else.
I see it as well on both w2k and linux. I was running with all cache disabled before, it makes it easier to spot some intemittnet bugs sorry for any confusion
I am continuing to see this behavior with the latest (2001-07-13 08) linux trunk build.
This appears to be fixed in the Linux nightly build of 2001073106.