Re: save-excursion and the mark

["Andreas Röhler" <andreas.roehler@online.de>]

On 25.02.2015 13:35, Ivan Shmakov wrote:
>>>>>> Andreas Röhler <andreas.roehler@online.de> writes:
>>>>>> On 25.02.2015 03:59, Stefan Monnier wrote:
>
>   >>> See for example indent-code-ridigly.
>
>   >> Can you give more details about how this requires save&restore of
>   >> the mark (or its "active" status)?
>
>   > Think a shifting code by TAB in python-mode.
>
> 	FWIW, python.el doesn’t seem to use the mark.
>
>   > Mark might exist somewhere in line. Shifting will use its own region
>   > – but original region should be restored – if any.  Shifting by
>   > itself should not create a region nor clutter the mark-ring.
>
> 	There seem to be a misunderstanding; the notion of /Emacs Lisp/
> 	region is different to that of /Emacs/ region, the former being
> 	(more or less) a way of saying “from here to there”, and
> 	does /not/ (generally) involve either mark or point.
>
> 	For instance, indent-code-ridigly can be used from Lisp like:
>
>     (indent-code-ridigly 13 37)
>
> 	Or perhaps:
>
>     (when-let ((here  (re-search-backward "START" nil t))
>                (there (re-search-forward  "END"   nil t)))
>       (indent-code-ridigly here there))
>
> 	With the common idiom being:
>
>     (let ((save (point)))
>       ; Move point forward, by some amount.
>       (indent-code-ridigly save (point)))
>
> 	As should be obvious, the mark is /not/ involved in either case.
> 	And, arguably, it /should not/ be.
>
> […]
>
>   > commit b1d6ddd44614c84746f5ee494e1f29cd9be8a2d8
>
>   > Notable changes to goto-line, i. e. calling push-mark.
>
> 	That makes a point, indeed: there’s a class of Emacs commands
> 	intended first and foremost for interactive use.  Naturally,
> 	goto-line is one of them; to quote its docstring:
>
> This function is for interactive use only;
> in Lisp code use `forward-line' instead.
>
> 	The proposed change to save-excursion would expose the bugs in
> 	the code which uses such commands inside of save-excursion,
> 	which now masks the unfortunate side-effects of such (mis)use.
>

Okay, thanks.

Andreas