Okay, this is a bit of a stumper

[annathepiper]

I'm seeing a discrepancy in what I see in some of our Japanese pages when I do a View Source on it on the QA version of our site vs. the prod version. Anybody want to help me suss this out?

So I go in IE 7 to the QA version of the page. I do a "View Source" on it and the page's source code pops up in Notepad as expected. I see a bunch of garbled characters that are supposed to be the Japanese characters.

I do this exact same thing on the production version of the page, and I see the actual Japanese characters. To wit, buh?

This therefore can't be an issue with Notepad failing to display the Japanese characters correctly. It can do it with the production page just fine. But what I don't get is why Notepad on the exact same computer shows me garbled characters for one page and Japanese characters for another one.

Note also that the QA version of the page displays the Japanese characters in the source code just fine if I look at it in Firefox.

Anybody got any theories?

Comments

[wingedelf.livejournal.com]

Is the page encoding set properly in the test environment?

[annathepiper]

As far as I know! The configuration files say UTF-8 in both QA and prod.

[mdlbear]

This was my first thought, too. Take a look at the HTTP response headers and see what they say for the encoding.

If the site's being served from Apache, check the .htaccess files as well as the site and host-wide configuration files. If there's browser-detection code on the server side, check that.

[annathepiper]

Headers have already been checked. But not Apache configuration files. Thanks!

Look at the HTTP headers, for sure.

[zeekar.livejournal.com]

Firefox has extensions that make it easy, or you can use command-line tools like curl -I, or lwp-request -m head... Look at the Content-Type: header, and see if it has a charset= parameter, and make sure they agree between the two versions. If they do, then check for a <meta http-equiv="content-type" content="..."> in the HTML, and compare those. If they're all the same, then I got bupkiss.

Re: Look at the HTTP headers, for sure.

[annathepiper]

I've already checked the HTTP headers (using Fiddler) and the meta tag, all of which agree as UTF8 in both QA and prod.

Re: Look at the HTTP headers, for sure.

[zeekar.livejournal.com]

Are the QA and prod versions on ttwo different web servers? The URL's otherwise the same? IE7 sometimes pays unwarranted attention to anything that looks like a file extension in a URL...

Re: Look at the HTTP headers, for sure.

[mdlbear]

Good point -- I've been burned by this, too. If the headers are the same, file extensions are a good place to look: IE often ignores the headers in favor of the extension. If there's no extension, IE usually does something completely braindead.

Re: Look at the HTTP headers, for sure.

[annathepiper]

See my comment to zeekar--the pages are on different web servers, and the only difference in the URL is in the domain name of the web servers, rather than the file name.

Re: Look at the HTTP headers, for sure.

[annathepiper]

They are different web servers, yes. And there is minor difference in the URLs, but it's at the beginning of the string in the domain name rather than at the end where the actual filename is.

"Anybody got any theories?"

[keethrax.livejournal.com]

Bunnies?

Re: "Anybody got any theories?"

[annathepiper]

They're AREN'T just cute like everybody supposes!

[fleetfootmike.livejournal.com]

Recommended reading, if you haven't: http://www.joelonsoftware.com/articles/Unicode.html

[annathepiper]

Thanks!

I actually eventually figured out the problem, or at least in terms of the immediate cause--though I'm not sure why the problem was happening. ;) There was a game on the QA site with some garbage characters in it and those were confusing Notepad when I looked at the source code.