I should give an example for this dual view mode that allows displaying both carets:
Imagine the edited logical text of one line is "...latinlatinlatin|ARABICARABICARABIC..." and the logical insertion point is exactly between the Latin and Arabic parts, the UBA-resolved display of the line is normally: || ...latinlatinlatin_...CIBARACIBARACIBARA^ || Where || marks the visible separation margins of the view, and _ is the RIGHT-pointing insertion caret at end of the UBA-resolved LTR span "...latin", and ^ is the LEFT-pointing insertion caret appearing at the begining of the UBA-resolved RTL span "ARABIC..." But the Arabic span is too long to fit fully in the edit view in order to whow the BEFORE logical context (...latin|) which should display here on the left, and the AFTER logical context (|ARABIC...) which should disply on the right. Then the view will be splitted as: || ...latin_ || ...CIBARA^ || Note that in this case of the LTR-to-RTL logical transition, both carets are shown to the RIGHT of their respective view. The BEFORE context is in the LEFT view, and the AFTER context is in the RIGHT view. If this was a RTL-to-LTR logical transition as in "...ARABICARABICARABIC|latinlatinlatin..." and if the display cannot fit the following view: || _latinlatinlatin...^CIBARACIBARACIBARA... || Then the dual view would become || _latin... || ^CIBARA... || And both carets would appear to the LEFT of their respective view in order to show enough context in each one. But the BEFORE context is now in the RIGHT view, and the AFTER context is in the LEFT view. In both cases: 1. The BEFORE or AFTER context with UBA-resolved LTR direction (latin) is shown in the LEFT view. In that view the displayed caret is always pointing to the LEFT 2. The BEFORE or AFTER context with UBA-resolved RTL direction (ARABIC) is shown in the RIGHT view. In that view the displayed caret is always pointing to the RIGHT 3. The other lines below the current line should be splitted in the same two views, in order to align their contents vertically with the visible content of the current line, as if there was no view split : we are seeing the same document, keepoing its logical margins, even if both side-margins of the document's page *may* eventually not be shown simultaneaously, as we can't fit the full page width in the available width for the total text input area.
