Hi Vadim
да I can understand your issue. In Czech, it is also more common to use spaces as the separator - thin spaces to be typographically precise. In WWW environments, the non-breaking space is usually used for for obvious reasons and AMOS does accept it. The actual input method depends on your keyboard configuration. I am GNU/Linux user so I use Compose+Space+Space to produce 0xC2 0xA0 bytes that represent U+00A0 in UTF-8 encoding.
But... Although non-breakable space would work fine for rendering numbers, yet another problem can emerge. The numerical question type uses the defined thousands separator when parsing and analysing students' responses, too. I did not test it but I am a bit afraid that if the non-breaking space is defined as the separator and students use the ordinary space, Moodle might consider their response as incorrect. Again, I did not test it and chances are that the question type authors were aware of this and the code copes with the situation well. If anybody can confirm it works or not, it would be very useful. If it does not work, people can easily face the situation that the student's response "10 000 000" (using ordinary spaces) is not the same as the correct "10 000 000" one (using non-breaking spaces).
So therefore I decided to use the comma as the separator for Czech even if it is less common (but still acceptable).
And yes, I think we will have to tweak the translator UI a bit and provide a way how we could declare that the translation should be defined as an empty string. We already had an issue with it - you may remember those delta size strings issue in TinyMCE, though there is a workaround of using zeros.
