bug#22819: 25.0.91; Don't try to indent region if the buffer is read-only

[Kaushal Modi]

On Sat, Aug 5, 2017, 8:10 AM Eli Zaretskii <eliz@gnu.org> wrote:

I see no problem in it, sorry.  And why was the user not aware of the
read-only status of the buffer to begin with?

Cockpit error, as you say later. It's easy to forget if you are working on a DVCS (like git) controlled file or CVCS (like SOS). 

 How

plausible is such a
scenario?

I resorted to writing this patch because it frustrated me quite a few times. 

 Are

we trying to change Emacs behavior to cater to a clear
cockpit error?

Isn't it better to alert user of an operation that will not succeed anyways beforehand, especially if the operation can be expensive like this one?

> >against veteran Emacs behavior regarding read->only text,
> >behavior that is present in several other >commands, and that AFAIR
> >resulted from some past discussions.
>
> This is the only one that provided me this surprise in about a decade of Emacs use. Which other commands
> do the text manipulation, and then check the buffer read-only status?

C-w, to name just one.

OK, it would be unnoticeable in that case as I have yet to see a kill operation that can take couple of minutes. 

IOW, a command could have useful side effects that are produced even
if the buffer is read-only and its text cannot be changed, thus
preventing the main effect of the command from happening.

> The flip question is: How common is a workflow, where a buffer is read-only, user does indentation, and is fine
> with seeing that error after the fact?

This goes both ways: if it's uncommon enough to be unimportant, then
changing the behavior is not important as well.

Would it be OK to open this up to emacs-devel to understand what workflows can break because of this change?

--

Kaushal Modi