Feed Sign in with OpenID OpenID

Simon Willison’s Weblog

Javascript character set screw-ups (via) Some browsers treat JavaScript files as having the same content-type as the page from which they are linked. This could cause problems with UTF-8 encoded JSON; the workaround is serving up ASCII with unicode escape sequences.

Posted 21st December 2006 at 3:20 pm

Tagged , ,

4 comments

  1. Well, he could add a charset="utf-8" attribute to the <script> tags. That fixes the issue in all browsers I've ever come across.

    I don't know about the XMLHTTPRequests (haven't tried) but I bet there's a similar property/method/attribute/thingy that we can use there.

    Már - 22nd December 2006 19:43 - #

  2. FWIW, Internet Explorer 7 fails both 2 and 3. It handles XMLHttpRequest ok. One could argue that Internet Explorer is the browser that's consistent here since browsers generally ignore the Content-Type header for script files. Apparently most still look at the charset parameter though.

    Anne van Kesteren - 24th December 2006 18:20 - #

  3. I've updated the web page with this, but I feel that I shouldn't have to know what charset the remote page is in. For those remote JSON APIs that are called by including a <script> tag, you'd need to make the charset of the remote page part of the public API. Not unreasonable, but (a) no-one will do it, and (b) it feel inelegant.

    Tom Insam - 26th December 2006 15:46 - #

  4. Oh, I totally agree with you. Just pointing out some facts.

    Anne van Kesteren - 26th December 2006 22:18 - #

Comments are closed.
A django site