Urs Hunkler

Another approach than to rewrite renderers to make classnames fit to different needs would be to implement a page postprocessing hook (Offered in several environments). If there where one it would be possible to change classnames with string replacements before the page is delivered to the browser. This approach is not ideal but I guess it would not take more time than to handle overridden renderers. These overriding tasks are quite time consuming in PHP.

Damyon, my impression is that one main reason why people would like to use CSS frameworks is the difficult situation to style Moodle - to fill the gaps.

When I remember right you had written that HTML/classnames and CSS are closely connected. Agreed. The best way out as I see it would be to clean up Moodle HTML/classnames and build consistent CSS. And not think about third party CSS frameworks which might need to fit on top of it - this approach will diffuse the task.

At Vimeo if you may be interested: This talk by Brad Frost was held at beyond tellerrand 2013, May 27 – 29, in Düsseldorf where he announced his brand new Pattern Lab: http://vimeo.com/channels/beyondtellerrand/67476280

PS: You may notice the short sequence at 50:20 in his talk

An interesting view on the discussion how to setup Moodle HTML and CSS I discovered when I wrote the postings. Moodle developers decided against a development framework and did the work to setup Moodle from scratch. If that is considered a good strategy for Moodle to be able to deliver the environment needed this approach may be right for the front-end too. And now with the front-end team Moodle HQ has the competences.

I think it is.

Damyon you start this discussion in the very moment I am fundamentally rethinking my Moodle theme work. With the switch to responsive layouts and multi device delivery a lot must change.

Not only needs the layout to adopt to different screens but also the page weight (KB/MB delivered) needs to  be reduced radically. And the content needs to be prepared fast on the server and rendered fast in the different browsers. So the key demands for Moodle pages which are capable to deliver an acceptable or even good user experience are responsive layouts and speed.

To implement responsive layouts is relative easy with Bootstrap. To work on the weight and speed is much more complicated. I know that the Moodle way to cascade themes by overriding CSS in several stages is not good for speed. To base the Moodle CSS on an existing framework is an easy starting point to get started immediately. But we buy in a lot of unused CSS.

A speedy and optimized solution you/we can get only by setting up the HTML structure and the CSS Moodle needs. Optimized for Moodle. One can steel everywhere and collect from existing approaches. Harry Roberts talked about "Architecting Scalable CSS" on the BeyondTellerand conference in May. His credo was "optimize". One slide I remember was "Each client needs their own Bootstrap" - I think the same is valid for Moodle.

When I look at the purecss.io/extend page today I like the way how Pure and parts of Bootstrap are combined. Moodle uses YUI - why not take purecss as a base and build the Moodle CSS on top. Use parts of Bootstrap where appropriate. And optimize, optimize, optimze.

By the way the pure "skin builder" looks interesting.