What is the difference between default and htmlparser2 tree adapters?

[domenic]

We are hoping to make parse5 the default in jsdom (jsdom/jsdom#818). I am curious whether we should just use the htmlparser2 adapter, and change none of our code, or if we should adopt the default parse5 format? Are there benefits?

[inikulin]

Hi, this is awesome news. default tree adapter was an original tree format for parse5 and it was designed to fit better into W3C-proposed HTML5 parser algorithm. Later I've implemented htmlparser2 tree adapter, so parse5 can be plugged into existing solutions that uses htmlparser2 tree format. There is a little performance penalty, but it's pretty insignificant. For jsdom I suggest to keep htmlparser2 tree format because given performance benefits will not cost effort spent on the tree format change. However, htmlparser2 tree adapter just mimics HTML Node's interfaces provided by htmlparser2, many other details (like named entity representation in text nodes, tree-hierarchy) may obviously differ from htmlparser2 due to completely different parsing algorithm (but it will almost same as in modern browsers).