Bug 15814 – fast/js/kde/encode_decode_uri.html fails

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Darin Adler]

The checked-in results include failures.

Because of a failure in fast/js/resources/js-test-pre.js this test used to pass when it was really failing.

[Darin Adler]

Created attachment 17015 [details]
patch

[Darin Adler]

<rdar://problem/5536644>

[Maciej Stachowiak]

Comment on attachment 17015 [details]
patch

r=me

[Alexey Proskuryakov]

+    return throwError(exec, URIError);

I think it would be good to add a string explaining the failure to throwError.

-    // Backwards BOM and U+FFFF should never appear in UTF-8 data.
-    if (c == 0xFFFE || c == 0xFFFF)
-      return -1;

FWIW, Firefox gives different result for these character on this test:

FAIL decodeURI(encodeURI(String.fromCharCode(65534))) should be . Was �.
FAIL decodeURI(encodeURI(String.fromCharCode(65535))) should be . Was �.

[Darin Adler]

(In reply to comment #4)
> FAIL decodeURI(encodeURI(String.fromCharCode(65534))) should be . Was
> �.
> FAIL decodeURI(encodeURI(String.fromCharCode(65535))) should be . Was
> �.

This behavior seems a bit strange and it's not mentioned by the standard. What do you think we should do?

[Darin Adler]

Committed revision 27406.

[Alexey Proskuryakov]

(In reply to comment #5)
> This behavior seems a bit strange and it's not mentioned by the standard. What
> do you think we should do?

I think the new behavior is correct, although I do not know why a comment in the removed code said that "Backwards BOM and U+FFFF should never appear in UTF-8 data." Of course, those are non-characters, but there are 64 more of those, and according to the Unicode FAQ, UTF encoding should round-trip those.

[Darin Adler]

(In reply to comment #7)
> I think the new behavior is correct, although I do not know why a comment in
> the removed code said that "Backwards BOM and U+FFFF should never appear in
> UTF-8 data." Of course, those are non-characters, but there are 64 more of
> those, and according to the Unicode FAQ, UTF encoding should round-trip those.

I wrote that comment. And I think I was wrong when I wrote it.