[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: tramp (2.3.3.26.1); tramp does not do a direct copy when it should
From: |
Michael Albinus |
Subject: |
Re: tramp (2.3.3.26.1); tramp does not do a direct copy when it should |
Date: |
Fri, 12 Jul 2019 12:56:08 +0200 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/27.0.50 (gnu/linux) |
Nicholas Gianniotis <address@hidden> writes:
Hi Nicholas,
> Steps to reproduce
>
> 1. visit a remote directory, (eg C-x C-f /scp:host:/some/folder)
>
> 2. You are now in `Dired by name' mode
>
> 3. Position the cursor on a test directory containing many subfiles
> (eg "test-A")
>
> 4. Type `R' (dired-do-rename)
>
> 5. Enter the new directory name (eg "test-B")
>
> 6. Tramp starts the rename, but does so by copying all files to local
> machine under new name, then copying back to remote host.
>
> 7. On a remote host over a slow connection the rename took 16 minutes.
I've tried to reproduce your scenario, in all my test cases Tramp
applies "mv -f ..." on the remote host. From your debug buffer, it is
not understandable why it calls "scp -p ..." twice.
So I need more information. Could you pls start Emacs like this:
--8<---------------cut here---------------start------------->8---
emacs -Q -l tramp -l tramp-sh
--8<---------------cut here---------------end--------------->8---
Then apply the following in the *scratch* buffer:
--8<---------------cut here---------------start------------->8---
(progn
(require 'trace)
(dolist (elt (all-completions "tramp-" obarray 'functionp))
(trace-function-background (intern elt)))
(untrace-function 'tramp-read-passwd))
--8<---------------cut here---------------end--------------->8---
Rerun your test scenario. You can stop traces then with
--8<---------------cut here---------------start------------->8---
M-x untrace-all
--8<---------------cut here---------------end--------------->8---
Send the resulting *trace-output* buffer. It shouldn't contain passwords
in clear text, but better cross-check.
> Thank you.
Best regards, Michael.