High performance text component

[Gaja Sutra]

I have a concern for localization support of FXML against the split of 
each paragraph in multiple FXML span tags.

By example, your first sample <p>Hello <b>Bold</b> <i>World</i></p>, can 
become in french <p>Bonjour <i>Monde</i> <b>Gras</b></p>, with different 
order of bold and italic styles. Like this case, FXML containing rich 
text will probably be separate for each language.

I understand DOM-like API for manipulation but I think it will be more 
complex to localize than some annotated string.

By example with a syntax like RTF/LaTeX <p>Hello \strong{Bold} 
\em{World}</p> and <p>Bonjour \em{Monde} \strong{Gras}</p>, you can 
localize only by substituting the string in the bundle, because your 
styles is not in FXML structure but only in the String containing text.

NB: In this case, your command annotating text is associated, by 
example, to a custom CSS pseudo-class:
p:strong {-fx-font-weight: bold;}
p:em {-fx-font-style: italic;}

NB: I know RTF/LaTeX syntax is not really beautiful. I am choosing this 
syntax only because special characters are not the same than XML and 
because ${...} is more frequently used for executing content (variable 
evaluation, etc.).