Re: [Nano-devel] problems when chunks are of variable length

[David Ramsey]

On Thu, Mar 30, 2017 at 12:52 PM, Benno Schulenberg
<address@hidden> wrote:
> The % editwincols formula will always yield the correct column: the
> correct /intended/ column if the relevant chunk were long enough.
> It's the same as in current git when the last chunk of a line does not
> reach until the column where we want to be: then we put the cursor
> after the last character of that chunk /but/ remember the
> target_column as the place we would like to be.

Okay.  I think we're talking about the same thing, just in slightly
different ways.  I just meant, if the line is broken early, that makes
the intended column a bit more difficult to reach depending on exactly
where the line is broken, but if it's not broken early, then, as you
say, the % editwincols formula will always yield the correct column.
(If I understand this correctly.)

> The difficult part (it seems to me) is that actual_x() cannot be
> computed simply from placewewant but has to be computed from the
> minimum of placewewant and the end of the current chunk.

Effectively, it adds a disconnect between placewewant and target_column.
placewewant can be past the end of the line, but target_column can't.
That's what my patch set has keep_before_softwrap_breakpoint() for; you
set placewewant to leftedge + target_column, then call
keep_before_softwrap_breakpoint() on target_column to get an
in-screen-range target_column, and then set current_x by calling
actual_x() not with placewewant, but with leftedge + the in-screen-range
target_column.