Re: [vile] Quoted motion: (un)bounding of rectangular changes

[Marc Simpson]

On Thu, Jun 15, 2017 at 5:37 AM, Paul Fox <address@hidden> wrote:
> marc wrote:
> > [snip]
>  > Agreed re: cw vs. cqwq, thanks for clarifying (I was treating change +
>  > sweep as a completely separate operation). Frankly, I'm finding quoted
>
> well, without thinking too hard about it, i'd say that the
> traditional "exceptions" that 'c' causes to word motions should probably
> be suppressed when doing quoted motions.  i probably didn't consider
> that case at the time -- i think i was picturing quoted motions as
> being more like:
>     cq<arrow><arrow><arrow><arrow><arrow><arrow>q
> since the concepts of "word", "next X character", etc don't feel
> as naturally applicable (to me) with rectangular selections.

Similar impression here—I've been considering cqwwq as:

0. change the following
1. [enter sweep mode]
2. step a word, then another (i.e., interactively move the cursor)
3. [end sweep mode and complete the change over the bounded region]

That is, my expectation in step (2) is for motion to behave as it does
in normal/visual mode with the change queued for after completion of a
selection; 'c' and 'w' are decoupled.

>  > motion a tad confusing—some of the issues raised are clearly due to
>  > user error.
>  >
>  > One more example:
>  >
>  > - cqwwq: only changes the first word
>  >   - The second word motion essentially behaves as l;
>  >   - cq2wq, the most direct analogue to normal vi, works as expected.
>
> that feels like a real bug.

Ack.

/M