bug#34908: Push mark in xref-push-marker-stack

[Juri Linkov <juri@linkov.net>]

>> I'm not sure where to call push-mark: closer to the command,
>> or closer to ring-insert.  It seems a suitable place for
>> push-mark is in xref-push-marker-stack as its name suggests.
>
> I'd rather have a more localized change, at least for now. And leave
> xref-push-marker-stack to only modify its own data structure.

Do you think other commands that use xref-push-marker-stack
won't need setting the mark?

> So how about this?
>
> diff --git a/lisp/progmodes/xref.el b/lisp/progmodes/xref.el
> index 6974d00048..015ea16f34 100644
> --- a/lisp/progmodes/xref.el
> +++ b/lisp/progmodes/xref.el
> @@ -797,6 +797,7 @@ xref--read-identifier-history
>  (defvar xref--read-pattern-history nil)
>
>  (defun xref--show-xrefs (xrefs display-action &optional always-show-list)
> +  (push-mark nil t)

push-mark is usually called closer to the end of the command.

>    (cond
>     ((and (not (cdr xrefs)) (not always-show-list))
>      (xref-push-marker-stack)

Some commands like e.g. ‘goto-line’ use such idiom:

  (or (region-active-p) (push-mark))

documented in its docstring as

  Prior to moving point, this function sets the mark (without
  activating it), unless Transient Mark mode is enabled and the
  mark is already active.