Themes

My thoughts are that the best approach for RTL is:

1. Moodle serving a single rtl.css or ltr.css file to the user, depending on the language used on the current page.

2. Moodle core and theme developers using rtl-flipping technology so that they only have to maintain a single theme file.

In the meantime, until that is all done, a potential workaround that lets us get most of the work done in parallel while waiting is to provide a setting in a theme that allows the admin to select ltr, rtl or both.

The ltr file is the main one that people work on. The rtl file is an automatically flipped version of that, possibly with some hand-crafted rtl-only tweaks appended.

The combined file is trickier, but my current approach would be to fuse the previous two files together and add [dir="ltr"] to the start of every line intended for ltr, and [dir="rtl"] to the start of every line intended for rtl. This basically means that the page would ignore half of the CSS based on which language direction it was using.

I believe the above should work, and I have a basic prototype, but I haven't had much time recently to test it out.

The immediate reaction to sending double Moodle's already quite large CSS to those sites that use both language directions is probably horror. But if you do some maths, with fairly rough numbers like 5% of Moodle users needing both RTL and LTR within a single theme and 10% of Moodle's current CSS being RTL specific you can see that:

current way = everyone gets a slightly larger CSS file = 1 x 1.1 = 110%
half way = most people get a slightly smaller CSS, some people get double = (.95 x 1) + (.05 x 2) = 105%
best way = everyone gets only the CSS they need = 1 x 1 = 100%

So overall the amount of CSS sent by Moodle could actually fall overall (depending on what the real numbers are) even though certain sites would be sending double. Of course, as mentioned above the real solution is to send only rtl for rtl pages and ltr for ltr pages. I'd guess this would be better even for sites that use both.

Generally, like most of the CSS it could just generally do with a tidy up e.g. part of the RTL CSS that is sent to every user is for the TinyMCE editor. And it's sent even if you're not using RTL, and even if you are using RTL but not using the TinyMCE editor. A few hundred extra lines of CSS here and there soon adds up.

OK...following my rant on the problems of the Essential Theme.

The real problem with RTL is...

1. it needs lots of browser support with Bi-Directional options which are slow in being developed
2. it needs lots of theme support with CSS specific selectors like .dir-rtl placed in the correct order relating to the body class in Moodle (not everyone knows this).
3. care needs to be taken when using CSS properties like paddings and margins, background-position, floats etc.
4. also care needs to be taken when using directional properties like direction: rtl as these can cause problems with punctuation forcing commas and full-stops, brackets etc,, to the wrong part of the sentence

These are only a few of the many problems with RTL in general, but would perhaps be an a great idea if something good could be done with Moodle for the RTL communities around the world. It has been a long time coming but we still need the support for all this in Browsers which is well supported in IE strangely enough. Did you know for instance that you can actually change the direction of the page in IE.  I think you can also do this in Mozilla too.

Splitting CSS up is one why or making a theme RTL specific is another.

Cleaning up Boostrap CSS might be better as there are lots of things in bootstrap.css that is never used.

If it were me I would advocate that Moodle design their own version of Bootstrap and call is MoodleStrap. In other words something that matches what Moodle is all about, that of a Modular Object-Oriented Dynamic Learning Environment