Bug 26371

Created attachment 31239 [details] First attempt Here is an enhancement for SVGLangSpace. Note that FF, Opera and Batik do not seem to know the js attributes xmllang and xmlspace at all, so it is hard to compare. Also really exceptions should be thrown when the attribute is set, but I think due to a bug that objC interfaces can't handle the idl setting of raises exception, that that may depend on another bug. I dont think this fixes a big issue but the old implementation wasted memory and processing time IMHO. All tests still pass, also because we do not really test these attributes except for one SVG text test. Cheers, Rob.

Comment on attachment 31239 [details] First attempt Hi Rob, nice patch, going in the right direction :-) > -void SVGElement::setXmlbase(const String& value, ExceptionCode&) > +void SVGElement::setXmlbase(const String&, ExceptionCode& ec) > { > - setAttribute(XMLNames::baseAttr, value); > + //setAttribute(XMLNames::baseAttr, value); > + ec = NO_MODIFICATION_ALLOWED_ERR; > } Is that change related? It contains commented code, hence I'm wondering. I guess you wanted to fix "attribute [ConvertNullToNullString] DOMString xmlbase setter raises(DOMException);" I'd rather commit this in a seperated patch with an associated testcase. > -void SVGLangSpace::setXmllang(const AtomicString& xmlLang) > +void SVGLangSpace::setXmllang(const AtomicString&/*, ExceptionCode &ec*/) > { > - m_lang = xmlLang; > + //ec = NO_MODIFICATION_ALLOWED_ERR; > } This either needs to throw an exception (then we need to correct the SVGLangSpace.idl file as well, fixing the commented out "setter raises..." lines). I guess this is what W3C says IIRC. If you go for that, we need a testcase. If you're on this anyway, I think it would make sense to go for that... > const AtomicString& SVGLangSpace::xmlspace() const > { > - if (!m_space) { > - DEFINE_STATIC_LOCAL(const AtomicString, defaultString, ("default")); > - return defaultString; > - } > - > - return m_space; > -} > - > -void SVGLangSpace::setXmlspace(const AtomicString& xmlSpace) > -{ > - m_space = xmlSpace; > + return contextElement()->getAttribute(XMLNames::spaceAttr); > } > > -bool SVGLangSpace::parseMappedAttribute(MappedAttribute* attr) > +void SVGLangSpace::setXmlspace(const AtomicString&/*, ExceptionCode& ec*/) > { > - if (attr->name().matches(XMLNames::langAttr)) { > - setXmllang(attr->value()); > - return true; > - } else if (attr->name().matches(XMLNames::spaceAttr)) { > - setXmlspace(attr->value()); > - return true; > - } > - > - return false; > + // ec = NO_MODIFICATION_ALLOWED_ERR; > } Same comment as above. Everything else looks fine. Because of lacking testcases setting r- to clear it out the review queue for now.

Created attachment 141628 [details] Patch

I checked, but on svg/dom/SVGStyleElement/script-tests/style-langspace.js Mozilla and Opera do not throw exceptions either and allow setting of the attributes, so I ignore the exception stuff for now.

Comment on attachment 141628 [details] Patch LGTM. Unclear to me why isKnownAttribute is needed for the early returns. It would also be OK to let execution continue and just fall off the end of parseMappedAttribute.

Comment on attachment 141628 [details] Patch Rejecting attachment 141628 [details] from commit-queue. Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 2 Last 500 characters of output: at 128 (offset -1 lines). patching file Source/WebCore/svg/SVGTitleElement.h patching file Source/WebCore/svg/SVGUseElement.cpp Hunk #1 FAILED at 158. 1 out of 1 hunk FAILED -- saving rejects to file Source/WebCore/svg/SVGUseElement.cpp.rej patching file Source/WebCore/svg/SVGUseElement.h Hunk #1 succeeded at 130 (offset 2 lines). Failed to run "[u'/mnt/git/webkit-commit-queue/Tools/Scripts/svn-apply', u'--force', u'--reviewer', u'Eric Seidel']" exit_code: 1 cwd: /mnt/git/webkit-commit-queue/ Full output: http://queues.webkit.org/results/13421215

This looks like a nice improvement, do you plan to fix it up and land it?

This code was refactored and removed.