Re: bug#34160: About commit bff64115a0ad081282e0f99305f41c8dd1917d67, bug#34160, json.el

[Tassilo Horn]

> The user referred to
> "http://git.savannah.gnu.org/cgit/emacs.git/tree/lisp/json.el#n740";,
> which doesn't look like Emacs 24?

Ah, might be.

> > Could you please reinstall the feature or describe why it is not
> > feasible to keep it?
>
> As the bug in question described -- pretty-printing a JSON region would
> silently delete everything but the first JSON object, which doesn't seem
> like optimal behaviour for a pretty-printing function.

Obviously not. :-)

> If there's a problem where point is moved unnecessarily, then that
> should be fixed, of course.  Do you have a test case?

It's not just moving point. replace-region-contents also keeps marks, text 
properties and fontification intact. So we should definitely be using it here.

The loop over all json objects in the region you've added is correct. It's 
just that I beg you to drop the delete-region / insert in favor of 
replace-region-contents.

json-read advances point until the end of the read json. This can be used 
to give the right region (not the complete region as I did) to the repeated 
replace-region-contents calls.

Feel free to give it a try. Otherwise I'll do it on the weekend.

For a test case for point keeping its position in the json: use my command 
from my original mail and an arbitrary json file and invoke it while point 
is somewhere inside the json object.

Bye,
 Tassilo