However, let me burn some karma this fine morning:
All of these layout libraries are, in my opinion, indicative of the fundamentally flawed layout model of HTML and CSS: it is overly complicated and can be extremely difficult to achieve bog-standard layouts (e.g. name-value pair forms) without lots of css work (in relative terms) and a deep understanding of the CSS layout system. For example, google "center a div on a page" and witness the carnage. I find the Swing layout managers, for example, far easier to work with in achieving a given UI that looks correct and resizes correctly as well. The client-side web has ignored many of the lessons of thick client development, much to its detriment IMO.
Same goes for mixing code and markup. When I put a button on a page, I want to put the code associated with that button in the definition of the button. Not in some random file tied to the button by a brittle and easily-duplicated id. Separating the markup and the programmatic logic is now orthodoxy, but it's a stupid orthodoxy: yes, yes, you don't want to put domain logic in your UI, but why the hell would you want to give up being able to quickly ascertain what a button does, even if you didn't create the UI?
Downvote away, kids.