Once bug 559303 is fixed, having another buffer in the tree builder becomes just a matter of object oriented design. However, the buffer in the tokenizer and the buffer in the tree builder are never needed at the same time, because text content tokenization doesn't overlap the tokenization of anything inside doctypes, comments or tags. Therefore, to save memory (and maybe to get a cache locality win if we are lucky), it would make sense to consolidate the buffer from the tree builder into the tokenizer buffer. This would be totally wrong from an OO design POV, of course.
You need to log in before you can comment on or make changes to this bug.