--- Begin Message ---
Subject: |
29.0.50; dired can't visit a local directory if another buffer visits an unreachable remote (TRAMP) directory |
Date: |
Thu, 24 Mar 2022 00:15:10 -0700 |
Hi. This sounds convoluted, but it's a real-life bug I hit repeatedly.
Recipe:
1. emacs -Q
2. C-x C-f /ssh:host:directory
Visit some remote directory via TRAMP. This opens a dired buffer
3. Unplug the network cable. Or walk away from wifi. Or turn off the
remote host. Optionally, do other stuff with emacs. The bug is hit if
the buffer in step 2 is open somewhere; doesn't have to be visible
4. C-x C-f /some/local/directory
One would expect step 4 to work just fine because it's a local
directory, so no TRAMP business should affect it, but it does. Step 4
will try to connect to the host in step 2. But that host is gone, so
it'll fail, and step 4 will fail too.
The problem is here:
(defun dired-find-buffer-nocreate (dirname &optional mode)
....
(let (found (blist dired-buffers))
...
(while blist
...
(with-current-buffer (cdr (car blist))
....
(expand-file-name
(if (consp dired-directory)
(car dired-directory)
dired-directory))))
Here dired is trying to figure out if we're already visiting the
requested directory. In doing so, it loops through all extant dired
buffers, and runs (expand-file-name) on each one. Running
(expand-file-name) on a remote directory will try to contact the host.
I'm not attaching a patch because I'm not 100% sure what's appropriate.
I guess we want to replace
(expand-file-name dired-directory)
with something like this pseudo-code
(concat (tramp-host dired-directory)
(expand-file-name (tramp-filename dired-directory)))
Except, thinking about edge cases: making sure this works with/without
TRAMP, making sure relative directories work right, etc, etc.
Do we already have a function that does this?
Thanks!
--- End Message ---
--- Begin Message ---
Subject: |
Re: bug#54542: 29.0.50; dired can't visit a local directory if another buffer visits an unreachable remote (TRAMP) directory |
Date: |
Sat, 26 Mar 2022 11:57:59 +0100 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/29.0.50 (gnu/linux) |
Version: 29.1
Dima Kogan <dima@secretsauce.net> writes:
Hi Dima,
>> We have already variable non-essential, which shall help for
>> non-existing / broken connections. While I could reproduce your
>> initial recipe locally, the attached patch cures it for me. Could you,
>> please, test?
>
> Hi Michael. Your patch fixes the issue. Thanks for looking into this!
Thanks for the feedback. I've committed this to Emacs master, closing
the bug.
Best regards, Michael.
--- End Message ---