812621 - "ASSERTION: Got the wrong answer from the custom method call" (nsIDOMHTMLElement_GetInnerHTML)

Status: NEW

(Core :: DOM: Core & HTML, defect, P5)

Description

[Jesse Ruderman]

Attachment: testcase

###!!! ASSERTION: Got the wrong answer from the custom method call!: 'NS_SUCCEEDED(debug_rv) == NS_SUCCEEDED(rv) && xpc_qsSameResult(debug_result, result)', file dom_quickstubs.cpp, line 8955

Comment 1

[Jesse Ruderman]

Attachment: stack+

Comment 2

[Jesse Ruderman]

Same problem with nsIDOMHTMLElement_GetOuterHTML, I think.

Comment 3

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

So, we call into GetMarkup, which calls GetContentType; this returns an empty string. The first time around, we don't have a cached encoder, so we try to create one by

      do_CreateInstance(PromiseFlatCString(
        nsDependentCString(NS_DOC_ENCODER_CONTRACTID_BASE) +
        [ empty string ]
      ).get());

which fails; we then do

    // This could be some type for which we create a synthetic document.  Try
    // again as XML
    contentType.AssignLiteral("application/xml");
    docEncoder = do_CreateInstance(NS_DOC_ENCODER_CONTRACTID_BASE "application/xml");

and are on our way. The second time around, though, we *do* have a cached encoder, so we skip the first if... the second if... and we leave contentType empty. Then we call into nsDocumentEncoder::EncodeToString, which does

    nsAutoCString progId(NS_CONTENTSERIALIZER_CONTRACTID_PREFIX);
    AppendUTF16toUTF8(mMimeType, progId);

    mSerializer = do_CreateInstance(progId.get());
    NS_ENSURE_TRUE(mSerializer, NS_ERROR_NOT_IMPLEMENTED);

with mMimeType set to the contentType from GetMarkup (which we'd passed to nsDocumentEncoder::NativeInit), i.e., the empty string. Unfortunately, we can't create an instance of "@mozilla.org/layout/contentserializer;1?mimetype=", and we bail out with NS_ERROR_NOT_IMPLEMENTED. GetMarkup passes it through to quickstubs, and we assert because NS_SUCCEEDED(rv) is true, and NS_SUCCEEDED(debug_rv) is false.

Which brings me to... Smaug and bug 574089?

Comment 4

[Henri Sivonen (:hsivonen)]

(In reply to :Ms2ger from comment #3)
> So, we call into GetMarkup, which calls GetContentType; this returns an
> empty string. The first time around, we don't have a cached encoder, so we
> try to create one by
> 
>       do_CreateInstance(PromiseFlatCString(
>         nsDependentCString(NS_DOC_ENCODER_CONTRACTID_BASE) +
>         [ empty string ]
>       ).get());

Why don't we set a content type for synthetic documents?

Why do we need the content type for innerHTML? Shouldn't we check the document's mIsRegularHTML and use the fast HTML serializer if true and always the same implementation of nsIDocumentEncoder (either always XML serializer or always XHTML serializer) if false?

Comment 5

[Firefox Bug Husbandry Bot]

https://bugzilla.mozilla.org/show_bug.cgi?id=1472046

Move all DOM bugs that haven’t been updated in more than 3 years and has no one currently assigned to P5.

If you have questions, please contact :mdaly.