bug#81467: [PATCH] Make bookmark-jump work with other-window/frame/tab-prefix

Stéphane Marks <[email protected]> Fri, 31 Jul 2026 17:38:02 +0200
Newsgroups gmane.emacs.bugs
Message-ID <CAN+1HbrFVEwFq3qytB=UFkR_BMVQ8-aHEbNOPbLrRg98e2Gp9A@mail.gmail.com>
On Fri, Jul 31, 2026 at 11:20 AM Stéphane Marks <[email protected]> wrote:

> On Fri, Jul 31, 2026 at 10:48 AM Stéphane Marks <[email protected]>
> wrote:
>
>> On Fri, Jul 31, 2026 at 10:35 AM Stefan Monnier <[email protected]>
>> wrote:
>>
>>> > 1. A bookmark handler 'shell-bookmark-jump' calls the function 'shell',
>>> > because for the authors it was the simplest way to implement the
>>> handler.
>>>
>>> And that's the part that needs fixing.
>>> E.g. by adding a `shell-no-select`.
>>>
>>
>> I'm happy to prepare a patch for that and at least a demonstration of how
>> I see using the bookmark-inhibit handler property that we'd assign to the
>> bufferlo handlers and bypass save-window-excursion, et.al.  Would these
>> be for emacs-31 or leave it alone for now and focus on master?
>>
>
> Thinking more about shell-bookmark-jump, if the default changes to
> shell-no-select, it would break bufferlo bookmarks that contain shell
> bookmarks.  Bufferlo currently passes #'ignore as display-function (via the
> advice I discussed earlier).  We'd need to come up with a way not to break
> those kinds of use cases.
>

Thinking even more, shell-bookmark-jump selecting or not will not
negatively impact bufferlo.  Each bufferlo tab handler which contains
things like shell bookmarks, invokes each bookmark's handler directly,
bypassing --jump-via's assumptions, and then restores the tab's window
configuration which can't bypass the originating --jump-via which may have
come from a bookmark-jump command (a somewhat nasty hack was written to
avoid the new save-window-excursion call which).  Top-level bufferlo frame
and tab bookmarks would still want bookmark-inhibit display-function
treatment for themselves.