Re: multi-region.el --- Mapping commands over multiple active regions.

Re: multi-region.el --- Mapping commands over multiple active regions.

[Richard M. Stallman]

I was going through old mail and came across your posting
of multi-region.el, last January.

Can someone imagine a way to integrate this into Emacs, so that
ordinary Emacs commands, which now operate on the region, could
operate on multi-regions when the user has specified multiple regions?

Here's one idea.  Suppose that region-beginning returned a list of
region beginnings, and region-end returned a list of region ends.
That would be a clean way to tell a command that there are
multi-regions in effect.

This could be done for (interactive "R"), whereas use
of (interactive "r") would simply repeat the command on each
sub-region in the list.

Now is not the time to install such a change, but we could think about
it now.

Re: multi-region.el --- Mapping commands over multiple active regions.

[Stefan Monnier]

> Here's one idea.  Suppose that region-beginning returned a list of
> region beginnings, and region-end returned a list of region ends.
> That would be a clean way to tell a command that there are
> multi-regions in effect.

> This could be done for (interactive "R"), whereas use
> of (interactive "r") would simply repeat the command on each
> sub-region in the list.

I completely agree.  As a matter of fact I suggested something along the
same lines some years ago on gnu.emacs.help (I especially remember the
(interative "r") automatically looping over all the sub-regions).


        Stefan

Re: multi-region.el --- Mapping commands over multiple active regions.

[Richard M. Stallman]

    I completely agree.  As a matter of fact I suggested something along the
    same lines some years ago on gnu.emacs.help (I especially remember the
    (interative "r") automatically looping over all the sub-regions).

Would you be interested in working on implementing something
for later installation?

Re: multi-region.el --- Mapping commands over multiple active regions.

[Mathias Dahl]

"Richard M. Stallman" <rms@gnu.org> writes:

> Here's one idea.  Suppose that region-beginning returned a list of
> region beginnings, and region-end returned a list of region ends.
> That would be a clean way to tell a command that there are
> multi-regions in effect.

Could there be risk here?

Suppose I have three regions:

r1: 10 to 20

r2: 30 to 40

r3: 50 to 60

If I understand your suggestion, `region-beginning' and `region-end',
respectively, would return the following values:

 (region-beginning) => (10 30 50)

 (region-end) => (20 40 60)

Could I, as a programmer, trust that the ordering of those values are
always in sync?

Maybe these functions are only helper functions, and when I need the
*real* end of a region, I will ask for it explicitly using the
beginning of the region as argument?

Re: multi-region.el --- Mapping commands over multiple active regions.

[Richard M. Stallman]

     (region-beginning) => (10 30 50)

     (region-end) => (20 40 60)

    Could I, as a programmer, trust that the ordering of those values are
    always in sync?

Sure, if you don't change anything between one call and the other.
They would both operate from the same data structure.

    Maybe these functions are only helper functions, and when I need the
    *real* end of a region, I will ask for it explicitly using the
    beginning of the region as argument?

Sorry, I do not understand the issue or the scenario.

Re: multi-region.el --- Mapping commands over multiple active regions.

[Kim F. Storm]

"Richard M. Stallman" <rms@gnu.org> writes:

>      (region-beginning) => (10 30 50)
>
>      (region-end) => (20 40 60)
>
>     Could I, as a programmer, trust that the ordering of those values are
>     always in sync?
>
> Sure, if you don't change anything between one call and the other.
> They would both operate from the same data structure.

Shouldn't those start/end positions be markers rather than simple numbers?
E.g. so that "delete-region" on a multi-region would work correctly...

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: multi-region.el --- Mapping commands over multiple active regions.

[Mathias Dahl]

"Richard M. Stallman" <rms@gnu.org> writes:

> Sure, if you don't change anything between one call and the other.
> They would both operate from the same data structure.
> ...
> 
>     Maybe these functions are only helper functions, and when I need the
>     *real* end of a region, I will ask for it explicitly using the
>     beginning of the region as argument?
>
> Sorry, I do not understand the issue or the scenario.

I cannot explain it better than that. I just got a bad feeling,
sorry. It's probably just me. :)

Re: multi-region.el --- Mapping commands over multiple active regions.

[Richard M. Stallman]

    Shouldn't those start/end positions be markers rather than simple numbers?

The underlying mechanism should use markers, so that they relocate.
And it makes sense to have a way to get at those markers themselves.
Perhaps that is the only interface we need.

We can work out those details as part of designing the feature,
if and when we do.]