599955 - Optimize DocumentEncoder

Status: RESOLVED FIXED

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

Description

[Olli Pettay [:smaug][[e-mail]]]

Attachment: a testcase

There are still several small things to optimize in DocumentEncoder.

The testcase is a kind of worst-case for gecko.
It shows problems in the DocumentEncoder recursion, SerializeNodeStart and
attribute entity conversion (which happens in serializer, and which this bug
is not about).

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

Attachment: wip

Something like this should help a bit.

Comment 2

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

You probably could replace some static_casts with AsElement in the code you're moving.

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Attachment: a bit simpler patch

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

So the patch adds a special case for the cases like innerHTML,
when it is enough to serialize contents of a node in a fast way.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 478874 [details] [diff] [review]
a bit simpler patch

>+++ b/content/base/src/nsDocumentEncoder.cpp
>@@ -359,29 +362,29 @@ nsDocumentEncoder::SerializeNodeStart(ns
>+  if (node->IsElement()) {
>+    nsIContent* originalElement =

Make that an Element*?

>+      aOriginalNode && aOriginalNode->IsElement() ?
>+        aOriginalNode->AsElement() : nsnull;

Maybe I should have added a static AsElement() instead, that cast null to null.  :(


>+    mSerializer->AppendElementStart(static_cast<nsIContent*>(node),
>+                                    originalElement, aStr);

And here you can pass node->AsElement() for the first arg, and then it will all Just Work when we change AppendElementStart to take Element.  ;)

With those nits, r=me.  This looks great!

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

Attachment: patch

This is part of getinnerhtml optimizations and would be great to get this
to ff4.

Comment 7

[Olli Pettay [:smaug][[e-mail]]]

http://hg.mozilla.org/mozilla-central/rev/c4f2c77eca93