Chris Graham

It definitely seems to be extends. I think there's some kind of combinatorial explosion going on here, but I don't have a couple of days to learn the parser /...

I should note that I've been profiling PHP a long time, and the numbers xdebug gives can be somewhat misleading. Function calls are pretty expensive, but the call cost itself...

I decided to take a closer look at this. The findMatch calls are complex, and called for every selector, for every extend. Let's say each rule block has 3 selectors,...

I measured the speed boost of the above properly. Compile time reduces by 40%, or 50% if you use trim (with '.# ') instead of preg_replace.

A quick look at the master link above suggests to me it would benefit from my optimisation also. I can't see any other fix they've made. I imagine in the...

It's not ideal as it will mess with column height balancing, but I think it is good enough.

It's also not ideal because it only looks at elements, not text nodes. So if you have a text node, then a removeiffirst element, it will get removed incorrectly. I'm...

Ergh, was planning to move code into the columnize function, but then I realised this is not going to be called for the last column, and I think cause other...

In fact... ``` $(this).find('table, thead, tbody, tfoot, colgroup, caption, label, legend, script, style, textarea, button, object, embed, tr, th, td, li, h1, h2, h3, h4, h5, h6, form').addClass('dontsplit'); ``` Of...

Micro optimizations really only make sense for code in the critical path (i.e. called a LOT during a single request), and I would very much doubt that hybridauth code is...