Alfredo Braunstein wrote:
Jean-Marc Lasgouttes wrote:
requires much more thinking to be sure of what it means. The original
version in particular can go outside of insets and continue from
there, which is not the case of the CursorSlice one.
If fact it is rather weird, and the "proof" is that we don't need this
feature at all in the code. Without that "feature" (and no caller code uses
it currently), the thing only moves on a flat cursorslice, I find
counterintuitive to have it on DocIterator.
+1.
Also concerning setFont, I am not sure we want to use CursorSlice as
parameter, although it makes sense technically. Things will get soon
annoying if for each function one has to wonder whether to use a full
cursor or a slice...
I have been advocating the contrary from the beginning ;-) Restricting the
data you have available make the code more robust, I find. IMO it's better
to use what's strictly necessary and no more... In that way you don't have
to guess if the thing acts recursively on all inner depths or just on the
same level: it's written on the function signature.
I agree with Alfredo. The Cursor code is horribly complicated because it
tries to do too many things at the same time. If a method only need a
CursorSlice, we should give it a CursorSlice.
Abdel.
