237566 - Remove nsIContent::ReplaceChildAt codepath

Status: RESOLVED FIXED

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

Description

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

The only place nsIContent::ReplaceChildAt is used is when someone calls
nsIDOMNode::ReplaceChild. However that function is called very rarly, both on
the web and in our chrome.

And while we in theory can save cycles by having a separate replace-path, the
place where we can save most cycles, the nsCSSFrameConstructor, doesn't really
take advantage of this. And it's very debatable if it should seing how rarly
it's used.

Also, since the codepath is so rarly used there's a big risk for bugs. Bug
193298 is one example.

Patch removing nsIContent::ReplaceChildAt and
nsIDocumentObserver::ContentReplaced comming up.

Comment 1

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

Attachment: patch to fix

Comment 2

[Boris Zbarsky [:bzbarsky]]

I very much doubt I'll be able to review this until I come back from my trip
(early April).

Comment 3

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

That's fine, the tree won't unfreeze until then anyway

Comment 4

[Johnny Stenback (:jst)]

Comment on attachment 143977 [details] [diff] [review]
patch to fix

sr=jst

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 143977 [details] [diff] [review]
patch to fix

>Index: content/base/src/nsGenericElement.cpp
>@@ -2883,22 +2845,15 @@ nsGenericElement::doReplaceChild(nsICont

>   nsCOMPtr<nsIContent> oldContent(do_QueryInterface(aOldChild, &res));
> 
>-  if (NS_FAILED(res)) {

Don't bother writing to res if you won't check it, ok?

r=bzbarsky with that change.  Looks good.

Comment 6

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

checked in, thanks for reviews!