Aquamacs 3.5 coming soon -- test now

[David Reitter]

What you can download on the build page is pretty close to what we intend to release as Aquamacs 3.5.

(As a signed binary, that is, so any associated issues will be addressed then.)

If you notice any severe new bugs, now would be the time to speak up.

--
http://aquamacs.org -- Aquamacs: Emacs on Mac OS X
http://aquamacs.org/donate -- Could we help you? Return the favor and support the Aquamacs Project!

[Lewis Hyatt]

It may not be quite what you had in mind for this announcement, but just thought I'd mention that I don't think anyone had time yet to fix the issue I reported with crashing when visiting a file that's already open.

https://groups.google.com/forum/#!topic/aquamacs-devel/ensH24d-P-U

I saw it was confirmed reproducible, but no fix yet. Anyway, if there's any chance it could be fixed, that would be amazing, since it's the only thing keeping Aquamacs from being perfect for my workflow. Thanks!

-lewis

[FIGUEROA-O'FARRILL Jose]

Hi,

dired-open-mac does not seem to work. I have it bound to “o” when in dired-mode, but it does nothing.

Not sure how to debug this.

Cheers,

José

> --
> You received this message because you are subscribed to the Google Groups "aquamacs-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to aquamacs-deve...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/aquamacs-devel/1476C99A-5C7C-4B40-BEEA-53B62D9499C6%40gmail.com.

Prof José Figueroa-O'Farrill
School of Mathematics
University of Edinburgh
PGP Key: 0x6A6BD529 (MIT PGP Key Server)

The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

[David Reitter]

Lewis,

I looked into this, and it seems that the patch below addresses the problem.

To explain, I think what happened is this. The one-buffer-one-frame package needs to hide frames sometimes rather than delete them. This is because Emacs expects at least one frame to be around at all times (receiving key events, for example, is tied to the selected frame).

The sequence on actions in your bug report hides a frame with buffer o1, because the o2 buffer is killed at one point. When that happens, the frame switches to showing o1, and that frame is hidden. Later, you ask to show file o1, and Aquamacs finds the hidden frame and determines that this should be made visible.

There are two bugs here. First, it should prefer the already visible frame and bring that up. That is now done with the patch.

Second, however, when a hidden frame is made visible, we have to call “make-frame-visible” because apparently that frame can be in some sort of zombie state, as you observed. I found this by selecting another frame, listing and finding the frame object with (frame-list) and (frame-parameters xx) and observing that “visibility” was nil, even though the frame was shown. So there was some inconsistency, and indeed, `make-frame-visible’ brought it back to life.

So, I’ll check the patch below in, and if you’d be so kind, please try out the new build.

David

--
http://aquamacs.org -- Aquamacs: Emacs on Mac OS X
http://aquamacs.org/donate -- Could we help you? Return the favor and support the Aquamacs Project!

diff --git a/aquamacs/src/site-lisp/one-buffer-one-frame.el b/aquamacs/src/site-lisp/one-buffer-one-frame.el
index afbe4db821..9f0a78a3f1 100644
--- a/aquamacs/src/site-lisp/one-buffer-one-frame.el
+++ b/aquamacs/src/site-lisp/one-buffer-one-frame.el
@@ -23,7 +23,8 @@
;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
;; Boston, MA 02111-1307, USA.

-;; Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2012, 2013, 2014 David Reitter
+;; Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2012, 2013, 2014,
+;; 2019: David Reitter

;; DESCRIPTION:
;;
@@ -406,9 +407,11 @@ obof-force-current-space
(switch t)
(window-to-select))
;; search for a window already displaying this buffer.
- (walk-windows
+ (dolist (include-hidden '(nil t))
+ (walk-windows
(lambda (w)
(when (and
+ (not window-to-select)
;; buffer is displayed in this window
(equal (window-buffer w) (get-bufobj (car args)))
;; Either window's frame *need not be* in the current space,
@@ -416,7 +419,7 @@ obof-force-current-space
;; or it *is* in the current space
(memq (window-frame w) (ns-visible-frame-list))))
(setq switch nil)
- (setq window-to-select w))) t t) ;) ;; t = include-hidden-frame (must be t)
+ (setq window-to-select w))) t include-hidden)) ;) ;; t = include-hidden-frame (must be t)
(if switch
;; Did *not* find a suitable window displaying the buffer.
(let ((same-window-regexps
@@ -453,7 +456,10 @@ obof-force-current-space
;; the next top-level event loop (assumption)
;; but because the normal switch-to-buffer does it right away
;; we should do it manually.
- (set-buffer (window-buffer window-to-select)))
+ (set-buffer (window-buffer window-to-select))
+ ;; Make sure frame is visible (can become unresponsive otherwise):
+ ;; https://groups.google.com/forum/#!topic/aquamacs-devel/ensH24d-P-U
+ (make-frame-visible (window-frame window-to-select)))
(unless ad-return-value (setq ad-return-value (current-buffer)))))
;; else: not one-buffer-one-frame
(if (or (ns-visible-frame-list)

> --
> You received this message because you are subscribed to the Google Groups "aquamacs-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to aquamacs-deve...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/aquamacs-devel/e22ac62e-5e2a-4e8a-9900-ea4b8206f28c%40googlegroups.com.

[Lewis Hyatt]

Thanks so much for looking into it, that's great. So far everything looks good to me, I am using this build now. Thanks again!

-Lewis

[David Reitter]

Hi Win, and everyone,

There will be a bit of a delay with this as we need to find out what’s causing a crash when you go full-screen and then close the frame (Command-W).  

I have applied a patch dated 10/7/2017 (by A. Third) but that caused more, strange effects.

If someone figures this one out, that would be amazing!

DR

--

http://aquamacs.org -- Aquamacs: Emacs on Mac OS X

http://aquamacs.org/donate -- Could we help you? Return the favor and support the Aquamacs Project!

-- 
You received this message because you are subscribed to the Google Groups "aquamacs-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to aquamacs-deve...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/aquamacs-devel/1476C99A-5C7C-4B40-BEEA-53B62D9499C6%40gmail.com.

[Win Treese]

On Aug 2, 2019, at 2:17 AM, David Reitter <david....@gmail.com> wrote:

Hi Win, and everyone,

There will be a bit of a delay with this as we need to find out what’s causing a crash when you go full-screen and then close the frame (Command-W).  
I have applied a patch dated 10/7/2017 (by A. Third) but that caused more, strange effects.

If someone figures this one out, that would be amazing!

I added this to the bug report (https://github.com/davidswelt/aquamacs-emacs/issues/160):

Was the patch from Emacs bug #26323? There's a follow-up discussion bug#26323: 25.1; Closing frames (in fullscreen) under Mac OS cause immediate crash.

I tried a simple build using the suggested compiler flag, but no joy. It looks like there are enough changes that have happened in the NS code that the most straightforward path is to look at this again after merging more recent core Emacs code, whether that's up through the latest Emacs 25.x or on to 26.

Should we continue any technical discussion about it there or on the list(s)?

 - Win