Re: [PATCH] Use prefix arg to control scope of org-narrow-to-subtree.

[Karl Fogel]

On 02 Dec 2019, Marco Wahl wrote:
>Karl Fogel <address@hidden> writes:
>> Since `widen' itself is already available via C-x n w, it might be
>> better to save a special flag value like that for some special
>> behavior that we (or someone else) might think of in the future.  I'm
>> pretty sure that anyone using `org-narrow-to-subtree' must also know
>> about the `widen' command, too.
>
>Okay, this sounds sound.
>
>Let me be a bit more explicit about my wish: I vote for the prefix arg 0
>to widen because this fits to the logic to view the whole file as level
>0 subtree.  With this perspective on the feature the effect of a numeric
>prefix argument is clear as day:
>
>...
>C-u 2 M-x org-narrow... => Narrow to the level-2 subtree containing point.
>C-u 1 M-x org-narrow... => Narrow to the level-1 subtree containing point.
>C-u 0 M-x org-narrow... => Narrow to the level-0 subtree containing point.
>
>The last line stands in opposition to the current behavior.
>
>C-u 0 M-x org-narrow... => Narrow to the level-1 subtree containing point.
>    ^                                          ^

Oh, yes, I get the logic, from a consistency standpoint.

My only concern is that a) it's unnecessary, because `widen' is available, and 
b) some day we might think of a better special meaning for C-u 0 (and in the 
meantime it could error instead of narrowing to the current level-1 subtree).

But I don't feel strongly about it either way.  Perhaps consistency is 
desirable here, although I tend to think that consistency is overrated in UI/UX 
in general -- the famous example of "`transpose-chars' at the end of a line" 
comes to mind.

Hmm, but on the other hand, your proposed consistency *would* be good if 
anyone's ever calling `org-narrow-to-subtree' from Lisp with an algorithmically 
calculated STEPS argument.  I can't imagine what kind of circumstance that 
would be, but if it's a general principle of Org Mode to consider "level 0" to 
be the entire buffer, then we should probably go with your proposed behavior.  
So I guess I'm tentatively +1...

Does anyone else have any thoughts on this?

Best regards,
-Karl