I tested how we parse all the labels in the encoding spec. Here are a bunch of labels that we don't support, but other browsers do. First, the spec says windows-949 should be an alias for euc-kr. WebKit recognizes it, IE doesn't. Second, here's a whole list of labels that the spec mandates and IE and WebKit recognize, but we don't: iso_8859-1:1978: windows-1252 (see bug 802030) iso_8859-2:1987: iso-8859-2 iso_8859-3:1988: iso-8859-3 iso_8859-4:1988: iso-8859-4 iso_8859-5:1988: iso-8859-5 iso_8859-6:1987: iso-8859-6 iso_8859-7:1987: iso-8859-7 iso_8859-8:1988: iso-8859-8 iso_8859-9:1989: windows-1254 (similar to bug 802030) csisolatin9, l9: iso-8859-15 (we recognize all the other csisolatin* and l* it defines, so I don't know why this one is different) cskoi8r, koi, koi8: koi8-r dos-874: windows-874 cn-big5: big5 sjis: shift_jis
Also, "logical" is a label for iso-8859-8-i. WebKit implements it per spec, while IE implements it as a label for iso-8859-8 (I don't think it implements iso-8859-8-i). I think that's it, but I might have missed some more.
I'm planning on submitting tests for this upstream at the W3C, maybe as early as today, and if I do they'll get synced back to us at some point. But I have to do some more work to get the tests usable (currently they rely on a PHP script), and I can't absolutely promise that will happen, so I won't mark in-testsuite+ yet. Tell me if you want tests for this, and if so, how to write them. Upstream I'm loading lots of iframes with <meta charset> in them, which is a bit inelegant . . .
Assignee: nobody → ayg
Status: NEW → ASSIGNED
Attachment #671777 - Flags: review?
Comment on attachment 671777 [details] [diff] [review] Patch Try: https://tbpl.mozilla.org/?tree=Try&rev=fa6ff5963936
Attachment #671777 - Flags: review? → review?(smontagu)
Comment on attachment 671777 [details] [diff] [review] Patch Review of attachment 671777 [details] [diff] [review]: ----------------------------------------------------------------- ::: intl/locale/src/charsetalias.properties @@ -134,5 @@ > iso8859-11=ISO-8859-11 > iso8859-13=ISO-8859-13 > iso8859-14=ISO-8859-14 > iso8859-15=ISO-8859-15 > -# Currently .properties cannot handle : in key I wonder how long this has not been true :)
Attachment #671777 - Flags: review?(smontagu) → review+
(In reply to :Aryeh Gregor from comment #2) > Tell me if you want tests for this, and if > so, how to write them. Upstream I'm loading lots of iframes with <meta > charset> in them, which is a bit inelegant . . . Please add the new aliases to the existing tests in intl/uconv/tests/unit/test-[en|de]code_*.js. (FTR, those use nsIScriptableUnicodeConverter; I don't know a better cross-browser way to test this stuff than iframes with <meta charset>, I'm afraid)
Done, but KOI8-R, ISO-8859-8-I, Shift_JIS, and Big5 didn't seem to have tests with that naming pattern, so I didn't do anything for them. https://hg.mozilla.org/integration/mozilla-inbound/rev/506a35df1d67
Status: ASSIGNED → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla19
You need to log in before you can comment on or make changes to this bug.