Forum Moderators: phranque
The issue:
Websites using a server that does on-the-fly conversion of cascading style sheets from text/plain to a gzip format for transmission(AddEncoding x-gzip .css?) leads to confusion in some users' browsers. The browser (IE6 for XPSP2 in our organisation) on some users' computers consistently download this sort of binary as text/plain.
The consequence of this is that on some computers, IE's rendering of certain sites simply "skips over" the scrambled-looking CSS text file ("scrambled" because it's binary labelled as plain/text?), making the site's formatting simply go all screwy in the best cases.
In other cases -- the worst cases -- the page doesn't open at all.. instead IE simply asks if you want to "download sitename.com".. Viewing the resultant download in a hex editor -- or indeed manually downloading an offending CSS file from any of the sites in question (URLs gleaned by viewing the pages' source codes) -- always reveals binary content starting with the GZIP header, 1F 8B 08.
Meanwhile, on _other_ computers in our fleet, the VERY SAME CSS file downloads as a true plain/text file (perfectly readable in NotePad). bwa!
Our fleet is a "single image" standard operating environment.. that is to say, the hardware is the same, the operating system is the same, the apps are the same (if the user doesn't add stuff). I've tried to find some obviously-related difference between the two states of machine (the "always works vs always doesn't work" states) but it's like searching for a needle in a haystack when the needle is made of hay. (a Terry Pratchett quote there.) I have ruled out any differences in filesets & registry entries deployed in the last four months of MS security updates. The updates all delivered the same on the working _and_ non-working desktops. bah.
Can any of you whiz kids provide a hint of how to proceed with this investigation. (Preferably one that doesn't mention the expected phrase "use Firefox™ instead," LoL.)
I can provide a coupla site URLs for the "best case & worst case" examples I mentioned, but I think I can only do that privately if the webmasterworld EULA i just agreed to is as strict on "namedropping" as it seems, heh. :o)
Thanks in advance! Hope someone knows what I'm going on about!
Have fun & mung beans,
Duck.
[edited by: NinjaDuck at 4:28 am (utc) on Oct. 7, 2007]
[edited by: jdMorgan at 2:03 pm (utc) on Oct. 8, 2007]
[edit reason] Formatting [/edit]
There is a fundamental difference between IE browsers and almost all others: Most browsers use the "Content-Type" server response header to determine how to handle content. However, IE prefers to 'examine' the actual content and take a best-guess as to what its content-type is. This can be a problem if the file is compressed and IE does not realize that before attempting to examine the content to determine its type.
Jim
