bug#5414: 23.1.91; /sys/devices files trigger loading tramp

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[jidanni]

Finding files like
/sys/devices/pci0000:00/0000:00:1e.0/0000:02:02.0
triggers "Loading tramp...done".
Maybe the regexps need to be tightened up.

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[Michael Albinus]

jidanni@jidanni.org writes:

> Finding files like
> /sys/devices/pci0000:00/0000:00:1e.0/0000:02:02.0
> triggers "Loading tramp...done".
> Maybe the regexps need to be tightened up.

I cannot reproduce it locally. Could you, please, start a fresh Emacs
via "emacs -Q", rerun the test, and create then a new message via "M-x
report-emacs-bug". This will show us the last keys you've entered.

You do not need to send the created message as-it-is; you might send it
to <5414@debbugs.gnu.org> in order to follow this bug's history.

Best regards, Michael.

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[jidanni]

>>>>> "MA" == Michael Albinus <michael.albinus@gmx.de> writes:

MA> I cannot reproduce it locally. Could you, please, start a fresh Emacs
MA> via "emacs -Q", rerun the test, and create then a new message via "M-x
MA> report-emacs-bug". This will show us the last keys you've entered.

$ set /tmp/sys/devices/pci0000:00/0000:00:1e.0/0000:02:02.0
$ install -D /dev/null $@
$ emacs -Q $@
C-x C-f
See how it says "loading tramp"?
In GNU Emacs 23.1.91.1 (i486-pc-linux-gnu, GTK+ Version 2.18.6)
(emacs-snapshot package, version 1:20100118-1)

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[Michael Albinus]

jidanni@jidanni.org writes:

>>>>>> "MA" == Michael Albinus <michael.albinus@gmx.de> writes:
>
> $ set /tmp/sys/devices/pci0000:00/0000:00:1e.0/0000:02:02.0
> $ install -D /dev/null $@
> $ emacs -Q $@
> C-x C-f
> See how it says "loading tramp"?

Negative. I have applied exactly the commands; Tramp is not loaded.

> In GNU Emacs 23.1.91.1 (i486-pc-linux-gnu, GTK+ Version 2.18.6)
> (emacs-snapshot package, version 1:20100118-1)

Which emacs-snapshot is this? Debian?

Best regards, Michael.

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[jidanni]

>>>>> "MA" == Michael Albinus <michael.albinus@gmx.de> writes:
MA> Negative. I have applied exactly the commands; Tramp is not loaded.
MA> Which emacs-snapshot is this? Debian?

$  apt-cache policy emacs-snapshot
emacs-snapshot:
  Installed: 1:20100118-1
  Candidate: 1:20100118-1
  Version table:
 *** 1:20100118-1 0
        500 http://emacs.orebokech.com sid/main Packages
        100 /var/lib/dpkg/status

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[Chong Yidong]

Here's a proper recipe for reproducing this (the previous recipe does
not do it, maybe it was untested):

 emacs -Q
 C-x C-f
 C-a C-k
 /sys/devices/pc [TAB]
    => Tramp will be loaded

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[jidanni]

Chong Yidong's recipe works for me, here with
$ ls /sys/devices/pc*
0000:00:00.0  0000:00:02.0  0000:00:02.5  0000:00:03.1  0000:00:0f.0   pci_bus  uevent
0000:00:01.0  0000:00:02.1  0000:00:03.0  0000:00:03.3  firmware_node  power

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[Michael Albinus]

Chong Yidong <cyd@stupidchicken.com> writes:

> Here's a proper recipe for reproducing this (the previous recipe does
> not do it, maybe it was untested):
>
>  emacs -Q
>  C-x C-f
>  C-a C-k
>  /sys/devices/pc [TAB]
>     => Tramp will be loaded

Thanks, I can reproduce it as well.

Tramp is loaded due to `rfn-eshadow-update-overlay'. In the loop
commented with "Binary search ...", it recomputes `mid', and calls
`rfn-eshadow-sifn-equal'. This function applies
`substitute-in-file-name' for a part of the the whole file name, which
looks like a remote file name, accidently. Bingo!

See the backtrace:

  tramp-file-name-handler(substitute-in-file-name "/pci0000:00/")
  substitute-in-file-name("/pci0000:00/")
  byte-code("ÁÂ\bd\"!‡" [pos substitute-in-file-name buffer-substring-no-properties] 4)
  byte-code("ÆÇ !È\b!É d\x19\x1a^[\x1c\vƒ;\0\v	=„'\0\f\vT\x1d‰\x1cÊË̏š*„;\0\v
=„‡\0\f\v\x1d‰\x1cÊË̏š*„‡\0
T	W„a\0	TdWƒ€\0\f	T\x1d‰\x1cÊË̏š*ƒ€\0	T\x12d‰\x11ƒ€\0
	\\Í¥\x13\f\v\x1d‰\x1cÊË̏š*ƒ{\0\v‰\x12‚<\0\v\x11‚;\0Î\bÉ 
#ˆÏÐ!,‡" [rfn-eshadow-overlay end start mid goal pos substitute-in-file-name minibuffer-contents overlay-end minibuffer-prompt-end nil (byte-code "ÁÂ\bd\"!‡" [pos substitute-in-file-name buffer-substring-no-properties] 4) ((error)) 2 move-overlay run-hooks rfn-eshadow-update-overlay-hook] 5)
  rfn-eshadow-update-overlay()
  run-hooks(post-command-hook)

I wouldn't regard it as a bug. One could argue, that rfn-shadow should
deactivate file name handlers for its work, but this might result in
collateral damages.

Best regards, Michael.

bug#5414: 23.1.91; /sys/devices files trigger loading tramp

[Chong Yidong]

Michael Albinus <michael.albinus@gmx.de> writes:

> Tramp is loaded due to `rfn-eshadow-update-overlay'. In the loop
> commented with "Binary search ...", it recomputes `mid', and calls
> `rfn-eshadow-sifn-equal'. This function applies
> `substitute-in-file-name' for a part of the the whole file name, which
> looks like a remote file name, accidently. Bingo!
>
> I wouldn't regard it as a bug. One could argue, that rfn-shadow should
> deactivate file name handlers for its work, but this might result in
> collateral damages.

Thanks for tracking this down.  I agree, there doesn't seem to be any
satisfactory way to get around this.  I'll close the bug.