[MonoDevelop] Managed Line Editor

[Kamil Skalski]

>  - Nearly all simple text-based operations (e.g. search/replace) would
> be slower and more cpu-intensive, since the editor would have to
> generate the text of the entire tree to perform them effectively.

Not quite.
Search/Replace? It is just a matter of good tree traversal function
(which should traverse it in order as it is printed to the
presentation layer). Note that at leaves of the tree I would still
keep the real strings.
But of course, it would be a little bit more expansive.

BUT, my guess it that 80% of used CPU / memory in modern IDEs is spent
on semantic analysis of text buffer contents - the simple "key
handing", "text deletion", "text insertion", "search" operations are
just some trivial iterations, which could run without problems on 386
era CPUs on even more complex structures we use now days. But ok,
that's all just my guesses. ;)

-- 
Kamil Skalski
http://nazgul.omega.pl