I volunteer to help with this :)

Thanks, Kparal. The bug related to this is:
https://bugzilla.redhat.com/show_bug.cgi?id=571900

Several cases have been drafted to divide global languages/keymap into groups and choose one representative language from each group: https://fedoraproject.org/wiki/Category:I18n_Installation

But I don't know what group Arabic belongs to and if they cover common circumstances. So any comment will be appreciated. Thanks!

I just included Arabic in my off-the-top-of-the-head list because it's a major non-Roman alphabet language - Arabic alphabet / input method is different again to CJK and Cyrillic conventions, AIUI. I'm certainly not an expert on these things, it may well be a good idea to ask i18n/l10n team.

Please add more comments on [https://fedorahosted.org/fedora-qa/ticket/83 ticket83]

I had a talk with Jens earlier through IRC, and he reviewed current tests. He thought the tests and their grouping idea are generally good so far, here I included his main comments.

{{{

<juhp> anyway overall I think it looks good so far
<juhp> you might need some adjustments after doing some real testing...
<juhp> also f13 has nice feature of enabling kbd switcher automatically in some cases

}}}

Next step I will do some real tests to adjust the steps. Regarding to the auto enabling switcher function, I don't have any physical keyboard other than qwerty. Can anyone who has other kinds of physical keyboard please test whether anaconda behaves different from using qwerty kbd during installation? I don't think so. But if yes, I will write a test about it.

Thanks.