Why does Firefox 18 ignore the specified character encoding for websites?
We are developing a page on our website that will have the page crawled and a newsletter generated and sent out to a mailing list. Many email packages default to character encoding of iso-8859-1 so we have set our character encoding to this on the page via the standard meta tag.
We have a problem on the newsletters that we had until now been unsuccessful to replicate. Though now I know why.... I have just discovered that in Firefox 18, the specified character encoding is being completely ignored. It is rendering the page in UTF-8 even though we specified ISO-8859-1. Firefox 3.6 however, renders the page with the proper encoding (thank god for keeping an old version for testing).
Can anyone explain why the new Firefox is completely ignoring the meta tag? Both browsers are using the factory default (I even opened FF18 in safe mode)...
Alle antwoorden (3)
Firefox ignores data set via a meta tag if the server already sends such data via the HTTP response headers and that may be happening in this case.
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
The only difference should be characters in the 0x80-0x9f range and it shouldn't be too difficult to generate such a document as UTF-8, so it would be best to switch to UTF-8 to prevent such issues.
Thanks for letting me know that Firefox 18 ignores everything but the server headers... but it doesn't help me much. Our website is in UFT-8... but this page is a newsletter, one that is crawled and saved into an email and sent out to a mailing list (by a third party newsletter program) and many email readers use ISO-8859-1 hence why we want to have the page rendered in that encoding so that we can actually test the newsletter properly. We can't test through the third party software as our testing environment is behind a firewall, and you can't change the server headers for a single page... hence the meta tag.
If you explicitly choose to render a page in a specific encoding, that shouldn't be ignored by the browser. It's not a big deal, but now every time we make a code change in our test environment and reload the page we have to force the encoding manually in the browser which is a pain.
The problem is, the newsletter is already live and we have some users complaining because some characters aren't displaying properly in their email packages (Entourage for Mac is one of them), all our testing (which is encoding using UTF-8) looks fine.
The only way to avoid this would be to use a different file extension for this page (i.e. not .html) and either make the server send no content type or the iso-8859-1 content type for this file extension.