bug#23794: Emacs 25.0.94: Patch to make sort-lines respect visible lines (fairly urgent)

[Eli Zaretskii]

> From: Robert Weiner <rsw@gnu.org>
> Date: Sat, 18 Jun 2016 13:42:01 -0400
> Cc: 23794@debbugs.gnu.org
> 
>  I don't think we can make such a backward-incompatible change without
>  (a) an entry in NEWS,
> 
> I am willing to write this.

Thanks.

>  (b) suitable changes in the manual(s), and
> 
> I looked at both the Emacs and the Elisp manuals and the only change 
> necessary would be in the elisp
> manual where the full code for sort-lines is shown, so that would be a simple 
> replace.

I think the fact that sorting ignores invisible text should be
prominently mentioned in both the user manual and the ELisp manual,
where the sort functions and commands are described.

>  (c) some way of getting back the old behavior (which could be by way
>  of having this new behavior as an optional one).
> 
> See below. 
> 
> For clarity, the original behavior of sort-lines is what the patch restores. 
> The backward-incompatibility to which
> you refer is then just an implementation error that occurred when switching 
> over to the overlay implementation
> of outlines as there was never any documentation that I can see that 
> suggested any behavior change. There
> certainly could be better documentation as to whether a 'line' refers to a 
> visible line, an invisible line or both
> but many functions do not delineate this. A major reason for making lines 
> invisible is so that they are not
> treated as regular lines when functions are applied to buffer text. Thus, 
> sort-lines should by default operate on
> visible lines. It could be extended or another function could be written to 
> operate on invisible lines as well, e.g.
> sort-invisible-lines and an alias could be made to sort-lines to be called 
> sort-visible-lines. All of this in the
> future. The only thing I am suggesting for right now is to restore the 
> original behavior. Note that if all lines are
> visible, the patch codes works as well. The issue is that when lines are 
> invisible the current code in Emacs
> does not work in a very useful way.

I think I already responded to this argument in my previous message.