390816 - setAttribute with ; doesn't work, NS_ERROR_DOM_INVALID_CHARACTER_ERR error

Status: RESOLVED INVALID

(Core :: DOM: Core & HTML, defect)

Description

[Martijn Wargers (dead)]

Attachment: testcase

See testcase, I get a js error with that testcase, because .setAttribute(';',';') isn't allowed. It works fine in IE and Opera. 

Actual result:
<pre id="result" style="white-space: -moz-pre-wrap; width: 600px;">[Exception... "String contains an invalid character"  code: "5" nsresult: "0x80530005 (NS_ERROR_DOM_INVALID_CHARACTER_ERR)"  location: "file:///C:/Documents%20and%20Settings/mw/Bureaublad/attributes2.htm Line: 6"]</pre>


Expected result:
<pre id=result style="WIDTH: 600px" ;=";"></pre>

What is even more string with Mozilla is that, if the element already has an attribute named ';', then it suddenly does work in Mozilla without errors.

Comment 1

[:Gavin Sharp [email: [e-mail]]]

I guess that attribute name is rejected by MOZ_XMLCheckQName, but it works if the attribute already exists because InternalGetExistingAttrNameFromQName finds it in that case and we don't rely on MOZ_XMLCheckQName to validate the name (via nsContentUtils::CheckQName).

Comment 2

[Boris Zbarsky [:bzbarsky]]

Sounds like MOZ_XMLCheckQName is buggy?  Or perhaps we shouldn't use it in HTML?

Comment 3

[Peter Van der Beken [:peterv]]

MOZ_XMLCheckQName is fine, ";" isn't a valid character for an QName in XML. Note that DOM Level 3 Core only talks about checking for XML name validity (http://www.w3.org/TR/2004/REC-DOM-Level-3-Core-20040407/core.html#ID-F68F082). I'm fine with switching to HTML5 name checking in HTML documents instead (see http://www.whatwg.org/specs/web-apps/current-work/#attributes1). Don't think this should have a very high priority though :-).

Comment 4

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

not blocking since we're not even sure this is a bug. Feel free to renominate if we actually determine that we do want to make a change, and we can find sites out there that does depend on it. Likely this is only going to be [wanted-1.9] at the most though.

Comment 6

[Peter Van der Beken [:peterv]]

Don't think ';' and ':' are the same, one is a valid name start character and the other is not.

Comment 7

[:Gavin Sharp [email: [e-mail]]]

(In reply to comment #6)
> Don't think ';' and ':' are the same, one is a valid name start character and
> the other is not.

OK. They're both rejected by MOZ_XMLCheckQName, though (for different reasons, http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/parser/expat/lib/moz_extensions.c&rev=1.11&mark=76,113#61 ).

Comment 8

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

AFAICT, Gecko/WebKit/Presto and the spec align here, while IE9 (in quirks mode, note) is too lenient.

Comment 9

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

(Thanks to KWierso for checking IE)

Comment 10

[Martijn Wargers (dead)]

Why not do the same as IE?

Comment 11

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Please send email to www-dom if you want this to change; I don't expect us to change this before the spec.

Comment 12

[Martijn Wargers (dead)]

I don't want the spec to change, I want Mozilla to change (perhaps).

Comment 13

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

If we believe the spec is wrong, we should be advocating to change it.

If we don't believe the spec is wrong, we should not be behaving differently.

Comment 14

[Martijn Wargers (dead)]

The spec should do what browsers are doing, not the other way around.