bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

[-- Attachment #1: Type: text/plain, Size: 834 bytes --]


For the following interactive shells:

* /bin/sh
* /usr/pkg/bin/zsh
* /usr/local/plan9/bin/rc

when "/usr/local/plan9/bin" is at the head of the PATH environment
variable and entry to GNU Emacs is by "emacs -Q", inside GNU Emacs the
dired command fails

# steps to reproduce behavior
* start by 'emacs -Q'
* list directory by 'C-x d'

# observed behavior
# minibuffer echo area says

> insert-directory: Listing directory failed but `access-file' worked

# expected behavior
# window displays listing of Dired, for example

'''
   /usr/X:
   total used in directory 96K available 72 GiB
   1966113 drwxr-xr-x  2 501 4.0K Jan 30 15:00 Articles
   1977600 drwxr-xr-x  3 501 4.0K Jan 31 07:11 Books
   1977546 drwxr-xr-x  2 501 4.0K Jan 26 20:00 Configurations
   1977542 -rw-r--r--  1 501 7.7K Jan 30 17:49 emacs-lisp.text
'''

-- 
vl

[-- Attachment #2: bug GNU Emacs NetBSD 9.2 Stable with pkgsrc-2021Q4 --]
[-- Type: text/plain, Size: 3707 bytes --]


In GNU Emacs 28.0.91 (build 1, x86_64--netbsd)
 of 2022-01-26 built on charlie
Repository revision: 3cf5af93293a4137550fcff623d6e2da25eed829
Repository branch: scratch/uni-unihan-readings
System Description: NetBSD charlie 9.2 NetBSD 9.2 (base) #3: Wed Jun 23 20:26:53 AEST 2021  uncle@charlie:/usr/src/sys/arch/amd64/compile/obj/base amd64

Configured using:
 'configure --without-all --without-x --without-compress-install
 --localstatedir=/var --disable-autodepend
 --prefix=/usr/X/Projects/emacs-28.0.91 --build=x86_64--netbsd
 --host=x86_64--netbsd --enable-option-checking=yes --without-pop
 --without-mailutils 'CFLAGS=-O2 -D_FORTIFY_SOURCE=2 -I/usr/include'
 'CPPFLAGS=-DTERMINFO -I/usr/include' 'LDFLAGS=-L/usr/lib -Wl,-R/usr/lib
 -Wl,-R/usr/pkg/lib''

Configured features:
JSON PDUMPER

Important settings:
  value of $LC_ALL: en_AU.UTF-8
  value of $LC_CTYPE: en_AU.UTF-8
  value of $LC_TIME: en_AU.UTF-8
  value of $LANG: en_AU.UTF-8
  locale-coding-system: utf-8-unix

Major mode: Text

Minor modes in effect:
  shell-dirtrack-mode: t
  electric-pair-mode: t
  icomplete-vertical-mode: t
  icomplete-mode: t
  tooltip-mode: t
  global-eldoc-mode: t
  show-paren-mode: t
  electric-indent-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  column-number-mode: t
  line-number-mode: t
  transient-mark-mode: t

Load-path shadows:
None found.

Features:
(shadow sort mail-extr emacsbug message rmc puny rfc822 mml mml-sec epa
derived epg rfc6068 epg-config gnus-util rmail tool-bar rmail-loaddefs
auth-source eieio eieio-core eieio-loaddefs password-cache json map
text-property-search time-date subr-x mm-decode mm-bodies mm-encode
mail-parse rfc2231 mailabbrev gmm-utils mailheader sendmail rfc2047
rfc2045 ietf-drums mm-util mail-prsvr mail-utils thingatpt vc-git
diff-mode easy-mmode vc-dispatcher bug-reference misearch multi-isearch
cl-seq em-unix em-term term disp-table shell ehelp em-script em-prompt
em-ls em-hist em-pred em-glob em-cmpl em-dirs esh-var pcomplete comint
ansi-color ring em-basic em-banner em-alias esh-mode eshell esh-cmd
esh-ext esh-opt esh-proc esh-io esh-arg esh-module esh-groups esh-util
make-mode seq sh-script smie executable view dired-aux regexp-opt dired
dired-loaddefs term/screen term/xterm xterm byte-opt bytecomp
byte-compile cconv elec-pair display-line-numbers edmacro kmacro
modus-vivendi-theme modus-themes cl-macs gv pcase icomplete cl-loaddefs
cl-lib cus-load iso-transl tooltip eldoc paren electric uniquify
ediff-hook vc-hooks lisp-float-type elisp-mode tabulated-list replace
newcomment text-mode lisp-mode prog-mode register page tab-bar menu-bar
rfn-eshadow isearch easymenu timer select mouse jit-lock font-lock
syntax font-core term/tty-colors frame minibuffer cl-generic cham
georgian utf-8-lang misc-lang vietnamese tibetan thai tai-viet lao
korean japanese eucjp-ms cp51932 hebrew greek romanian slovak czech
european ethiopic indian cyrillic chinese composite emoji-zwj charscript
charprop case-table epa-hook jka-cmpr-hook help simple abbrev obarray
cl-preloaded nadvice button loaddefs faces cus-face macroexp files
window text-properties overlay sha1 md5 base64 format env code-pages
mule custom widget hashtable-print-readable backquote multi-tty
make-network-process emacs)

Memory information:
((conses 16 186497 7854)
 (symbols 48 13883 3)
 (strings 32 36546 1801)
 (string-bytes 1 1986096)
 (vectors 16 17779)
 (vector-slots 8 214105 6432)
 (floats 8 87 272)
 (intervals 56 2102 0)
 (buffers 992 31))

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Wed, 2 Feb 2022 22:12:31 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> 
> For the following interactive shells:
> 
> * /bin/sh
> * /usr/pkg/bin/zsh
> * /usr/local/plan9/bin/rc
> 
> when "/usr/local/plan9/bin" is at the head of the PATH environment
> variable and entry to GNU Emacs is by "emacs -Q", inside GNU Emacs the
> dired command fails
> 
> # steps to reproduce behavior
> * start by 'emacs -Q'
> * list directory by 'C-x d'
> 
> # observed behavior
> # minibuffer echo area says
> 
> > insert-directory: Listing directory failed but `access-file' worked

I guess the version of 'ls' there is incompatible or something?  The
message means that 'ls' exited with a non-zero status for some reason.
Please investigate more what went wrong there and report here the
details, so we could decide whether we want to support "plan9/bin"
idiosyncrasies, whatever they are.  Stepping with Edebug through
insert-directory should help  you understand what's going on.

Also, please explain what you mean by "for the following interactive
shells".  How does the interactive shell enter this picture?  And what
is the value of shell-file-name in the session where "C-x d" fails?

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Eli Zaretskii <eliz@gnu.org> writes:

> I guess the version of 'ls' there is incompatible or something?  The
> message means that 'ls' exited with a non-zero status for some reason.

Yes, Plan 9 from User Space (which provides the ports of Plan 9
utilities to Unix-like systems) provides an ls binary that is
incompatible with what Emacs expects.

> Please investigate more what went wrong there and report here the
> details, so we could decide whether we want to support "plan9/bin"
> idiosyncrasies, whatever they are.

It would be a great deal of trouble to support their version of ls (and
indeed, their versions of many other common Unix utilities such as `rm'
and `cp'), so I think the answer is simply "don't run Emacs with
plan9/bin at the head of your PATH".

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: Van Ly <van.ly@sdf.org>,  53739@debbugs.gnu.org
> Date: Thu, 03 Feb 2022 16:01:51 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > I guess the version of 'ls' there is incompatible or something?  The
> > message means that 'ls' exited with a non-zero status for some reason.
> 
> Yes, Plan 9 from User Space (which provides the ports of Plan 9
> utilities to Unix-like systems) provides an ls binary that is
> incompatible with what Emacs expects.
> 
> > Please investigate more what went wrong there and report here the
> > details, so we could decide whether we want to support "plan9/bin"
> > idiosyncrasies, whatever they are.
> 
> It would be a great deal of trouble to support their version of ls (and
> indeed, their versions of many other common Unix utilities such as `rm'
> and `cp'), so I think the answer is simply "don't run Emacs with
> plan9/bin at the head of your PATH".

Can you tell more about the incompatibilities, perhaps by comparing
the output of their 'ls' with a compatible one?

Thanks.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Eli Zaretskii <eliz@gnu.org> writes:

> Can you tell more about the incompatibilities, perhaps by comparing
> the output of their 'ls' with a compatible one?

I can't remember the specifics, but IIRC their version of ls simply
prints out each file in the current working directory, without any
option to display permission bits, size, or ownership information.

Thanks.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: van.ly@sdf.org,  53739@debbugs.gnu.org
> Date: Thu, 03 Feb 2022 18:01:40 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Can you tell more about the incompatibilities, perhaps by comparing
> > the output of their 'ls' with a compatible one?
> 
> I can't remember the specifics, but IIRC their version of ls simply
> prints out each file in the current working directory, without any
> option to display permission bits, size, or ownership information.

You are saying that "ls -l" doesn't honor the -l option?

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Eli Zaretskii <eliz@gnu.org> writes:

>> I can't remember the specifics, but IIRC their version of ls simply
>> prints out each file in the current working directory, without any
>> option to display permission bits, size, or ownership information.

> You are saying that "ls -l" doesn't honor the -l option?

There isn't an -l option, IIRC.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: van.ly@sdf.org,  53739@debbugs.gnu.org
> Date: Thu, 03 Feb 2022 19:02:57 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> I can't remember the specifics, but IIRC their version of ls simply
> >> prints out each file in the current working directory, without any
> >> option to display permission bits, size, or ownership information.
> 
> > You are saying that "ls -l" doesn't honor the -l option?
> 
> There isn't an -l option, IIRC.

OK.  Do they have a replacement program that behaves like "ls -l"?
Maybe 'dir' or somesuch?  Then the solution would be to give a
different value to insert-directory-program on that platform.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Eli Zaretskii <eliz@gnu.org> writes:

> OK.  Do they have a replacement program that behaves like "ls -l"?
> Maybe 'dir' or somesuch?  Then the solution would be to give a
> different value to insert-directory-program on that platform.

I don't know.

But it's not really a special platform, just NetBSD with an IMSHO
misconfigured path full of utilities ported from another very niche
operating system that we don't yet support.

Thanks.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Thu, 3 Feb 2022, Po Lu wrote:

>
> But it's not really a special platform, just NetBSD with an IMSHO
> misconfigured path full of utilities ported from another very niche
> operating system that we don't yet support.
>

IIRC I saw mentioned Emacs was ported to Plan9.  A basic difference 
between UNIX and Plan9 is the obsolescense of the TTY.

=> https://plan9.io/
=> https://p9f.org/

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Thu, 3 Feb 2022 13:20:58 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: Eli Zaretskii <eliz@gnu.org>, 53739@debbugs.gnu.org
> 
> On Thu, 3 Feb 2022, Po Lu wrote:
> 
> >
> > But it's not really a special platform, just NetBSD with an IMSHO
> > misconfigured path full of utilities ported from another very niche
> > operating system that we don't yet support.
> >
> 
> IIRC I saw mentioned Emacs was ported to Plan9.

So maybe you should use that port.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

[-- Attachment #1: Type: text/plain, Size: 466 bytes --]

On Thu, 3 Feb 2022, Po Lu wrote:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> Can you tell more about the incompatibilities, perhaps by comparing
>> the output of their 'ls' with a compatible one?
>
> I can't remember the specifics, but IIRC their version of ls simply
> prints out each file in the current working directory, without any
> option to display permission bits, size, or ownership information.
>

The manpage for plan9's ls command is attached.

-- 
vl

[-- Attachment #2: Manpage for ls command --]
[-- Type: text/plain, Size: 3919 bytes --]


     LS(1)                                                       LS(1)

     NAME
          ls, lc - list contents of directory

     SYNOPSIS
          ls [ -dlmnpqrstuFQ ] name ...

          lc [ -dlmnpqrstuFQ ] name ...

     DESCRIPTION
          For each directory argument, ls lists the contents of the
          directory; for each file argument, ls repeats its name and
          any other information requested.  When no argument is given,
          the current directory is listed.  By default, the output is
          sorted alphabetically by name.

          Lc is the same as ls, but sets the -p option and pipes the
          output through mc(1).

          There are a number of options:

          -d   If argument is a directory, list it, not its contents.

          -l   List in long format, giving mode (see below), file sys-
               tem type (e.g., for devices, the # code letter that
               names it; see intro(3)), the instance or subdevice num-
               ber, owner, group, size in bytes, and time of last mod-
               ification for each file.

          -m   List the name of the user who most recently modified
               the file.

          -n   Don't sort the listing.

          -p   Print only the final path element of each file name.

          -q   List the qid (see stat(3)) of each file; the printed
               fields are in the order path, version, and type.

          -r   Reverse the order of sort.

          -s   Give size in Kbytes for each entry.

          -t   Sort by time modified (latest first) instead of by
               name.

          -u   Under -t sort by time of last access; under -l print
               time of last access.

          -F   Add the character / after all directory names and the
               character * after all executable files.

     Page 1                       Plan 9              (printed 2/4/22)

     LS(1)                                                       LS(1)

          -L   Print the character t before each file if it has the
               temporary flag set, and - otherwise.

          -Q   By default, printed file names are quoted if they con-
               tain characters special to rc(1).  The -Q flag disables
               this behavior.

          The mode printed under the -l option contains 11 characters,
          interpreted as follows: the first character is

          d    if the entry is a directory;

          a    if the entry is an append-only file;

          D    if the entry is a Unix device;

          L    if the entry is a symbolic link;

          P    if the entry is a named pipe;

          S    if the entry is a socket;

          -    if the entry is a plain file.

          The next letter is l if the file is exclusive access (one
          writer or reader at a time).

          The last 9 characters are interpreted as three sets of three
          bits each.  The first set refers to owner permissions; the
          next to permissions to others in the same user-group; and
          the last to all others.  Within each set the three charac-
          ters indicate permission respectively to read, to write, or
          to execute the file as a program.  For a directory, `exe-
          cute' permission is interpreted to mean permission to search
          the directory for a specified file.  The permissions are
          indicated as follows:

          r  if the file is readable;
          w  if the file is writable;
          x  if the file is executable;
          -  if none of the above permissions is granted.

     SOURCE
          /usr/local/plan9/src/cmd/ls.c
          /usr/local/plan9/bin/lc

     SEE ALSO
          stat(3), mc(1)

     Page 2                       Plan 9              (printed 2/4/22)

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Thu, 3 Feb 2022 13:12:07 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: Eli Zaretskii <eliz@gnu.org>, 53739@debbugs.gnu.org
> 
> > I can't remember the specifics, but IIRC their version of ls simply
> > prints out each file in the current working directory, without any
> > option to display permission bits, size, or ownership information.
> >
> 
> The manpage for plan9's ls command is attached.

This says the -l option is supported, so I'm not sure I understand
what is the cause of the problem.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Van Ly <van.ly@sdf.org> writes:

> The manpage for plan9's ls command is attached.

Please show some output from that ls command when you run it with the
"-l" option.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Fri, 4 Feb 2022, Po Lu wrote:

>
> Please show some output from that ls command when you run it with the
> "-l" option.
>

'''
; on NetBSD
% /usr/local/plan9/bin/ls -l
d-rwxr-xr-x M 0 van staff 0 Feb  1 02:37 Configurations
d-rwxr-xr-x M 0 van staff 0 Feb  3 21:08 Examples
d-rwxr-xr-x M 0 van staff 0 Jan 25 20:38 Projects

; on GNU/Linux
% /usr/local/plan9/bin/ls -l
d-rwxr-xr-x M 0 van dialout     0 Jan 31 16:15 .git
--rw-r--r-- M 0 van dialout    39 Jan 12 00:14 .gitignore
d-rwxr-xr-x M 0 van dialout     0 Jan 30 15:00 Articles
d-rwxr-xr-x M 0 van dialout     0 Jan 31 07:11 Books
d-rwxr-xr-x M 0 van dialout     0 Feb  2 13:19 Configurations
d-rwxr-xr-x M 0 van dialout     0 Jan 26 03:56 Examples
d-rwxr-xr-x M 0 van dialout     0 Jan 30 20:09 Letters
d-rwxr-xr-x M 0 van dialout     0 Jan 30 20:38 Notes
d-rwxr-xr-x M 0 van dialout     0 Jan 24 10:43 Projects
--rw-r--r-- M 0 van dialout  1513 Jan 31 16:15 README
--rw-r--r-- M 0 van dialout  8342 Feb  4 13:08 emacs-lisp.text
--rw-r--r-- M 0 van dialout   208 Jan 28 09:57 focs.text
--rw-r--r-- M 0 van dialout  2242 Jan 11 06:16 graphviz.text
--rw-r--r-- M 0 van dialout   140 Jan 31 08:09 ir-book.text
--rw-r--r-- M 0 van dialout   772 Jan 11 06:11 jau--zhou1--zau1.text
--rw-r--r-- M 0 van dialout  3605 Jan 28 19:55 modern-c.text
--rw-r--r-- M 0 van dialout 28133 Feb  4 09:50 scratch.text
--rw-r--r-- M 0 van dialout  1505 Jan 31 16:13 styleguide-eu-english.text
--rw-r--r-- M 0 van dialout  4096 Jan 30 19:30 typescript
'''

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Van Ly <van.ly@sdf.org> writes:

> % /usr/local/plan9/bin/ls -l
> d-rwxr-xr-x M 0 van staff 0 Feb  1 02:37 Configurations
> d-rwxr-xr-x M 0 van staff 0 Feb  3 21:08 Examples
> d-rwxr-xr-x M 0 van staff 0 Jan 25 20:38 Projects

That should probably work fine, so I have no idea what the problem might
be.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Fri, 4 Feb 2022, Po Lu wrote:

> Van Ly <van.ly@sdf.org> writes:
>
>> % /usr/local/plan9/bin/ls -l
>> d-rwxr-xr-x M 0 van staff 0 Feb  1 02:37 Configurations
>> d-rwxr-xr-x M 0 van staff 0 Feb  3 21:08 Examples
>> d-rwxr-xr-x M 0 van staff 0 Jan 25 20:38 Projects
>
> That should probably work fine, so I have no idea what the problem might
> be.
>

The problem is reproduced on GNU/Linux when plan9/bin is placed 
second after the head on PATH.

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Van Ly <van.ly@sdf.org> writes:

>> That should probably work fine, so I have no idea what the problem might
>> be.

> The problem is reproduced on GNU/Linux when plan9/bin is placed second
> after the head on PATH.

But the important question is which `ls' is run, not the order where
directories appear in PATH?

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: Eli Zaretskii <eliz@gnu.org>,  53739@debbugs.gnu.org
> Date: Fri, 04 Feb 2022 12:50:39 +0800
> 
> Van Ly <van.ly@sdf.org> writes:
> 
> > % /usr/local/plan9/bin/ls -l
> > d-rwxr-xr-x M 0 van staff 0 Feb  1 02:37 Configurations
> > d-rwxr-xr-x M 0 van staff 0 Feb  3 21:08 Examples
> > d-rwxr-xr-x M 0 van staff 0 Jan 25 20:38 Projects
> 
> That should probably work fine, so I have no idea what the problem might
> be.

The problem, as I explained up-thread, seems to be that 'ls' returns a
non-zero exit status, which insert-directory takes as a sign that the
directory listing didn't work.

So: what does "ls -l" return as exit status when using the plan9
variant of 'ls'?

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Fri, 4 Feb 2022, Eli Zaretskii wrote:

>
> The problem, as I explained up-thread, seems to be that 'ls' returns a
> non-zero exit status, which insert-directory takes as a sign that the
> directory listing didn't work.

The "9 man intro" page has this paragraph containing the "exit" keyword.

'''
      DIAGNOSTICS
           In Plan 9, a program's exit status is an arbitrary text
           string, while on Unix it is an integer.  Section (1) of this
           manual describes commands as though they exit with string
           statuses.  In fact, exiting with an empty status corresponds
           to exiting with status 0, and exiting with any non-empty
           string corresponds to exiting with status 1.  See exits(3).
'''

>
> So: what does "ls -l" return as exit status when using the plan9
> variant of 'ls'?
>

Starting 'emacs -Q' at the end of this typescript reproduces the problem.

'''
; on GNU/Linux
% echo $SHELL
/usr/bin/zsh
% PATH=/usr/local/plan9/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin; export PATH
% echo $PATH
/usr/van/.opam/cs3110-2-21fa/bin:/usr/local/plan9/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin
% which ls
/usr/local/plan9/bin/ls
% ls -l
--rw-r--r-- M 0 van dialout 46 Feb  4 18:22 emacs28
--rw-r--r-- M 0 van dialout 47 Feb  4 18:14 emacs28q
% echo $?
0
% /usr/local/plan9/bin/ls -l
--rw-r--r-- M 0 van dialout 46 Feb  4 18:22 emacs28
--rw-r--r-- M 0 van dialout 47 Feb  4 18:14 emacs28q
% echo $?
0

'''

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Fri, 4 Feb 2022, Van Ly wrote:

> On Fri, 4 Feb 2022, Eli Zaretskii wrote:
>
>> 
>> So: what does "ls -l" return as exit status when using the plan9
>> variant of 'ls'?
>> 
>

This also reproduces the problem.

'''
% echo $SHELL
/usr/bin/zsh
% export PATH=/usr/local/plan9/bin:/usr/X/Projects/emacs-2022-01-15/bin
% echo $PATH
/usr/local/plan9/bin:/usr/X/Projects/emacs-2022-01-15/bin
% which ls
/usr/local/plan9/bin/ls
% ls -l
--rw-r--r-- M 0 van dialout 46 Feb  4 18:22 emacs28
--rw-r--r-- M 0 van dialout 47 Feb  4 18:14 emacs28q
% echo $?
0
% /usr/local/plan9/bin/ls -l
--rw-r--r-- M 0 van dialout 46 Feb  4 18:22 emacs28
--rw-r--r-- M 0 van dialout 47 Feb  4 18:14 emacs28q
% echo $? 
0
% emacs -Q
'''

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Fri, 4 Feb 2022 09:26:28 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: Po Lu <luangruo@yahoo.com>, 53739@debbugs.gnu.org
> 
> On Fri, 4 Feb 2022, Van Ly wrote:
> 
> > On Fri, 4 Feb 2022, Eli Zaretskii wrote:
> >
> >> 
> >> So: what does "ls -l" return as exit status when using the plan9
> >> variant of 'ls'?
> >> 
> >
> 
> This also reproduces the problem.

I don't see what it reproduces, and more importantly, don't see how
this is relevant.  Please step through the code of insert-directory
and describe here what you see there when it reports the error.
Everything else is just noise, as far as investigation of this problem
is concerned.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

[-- Attachment #1: Type: text/plain, Size: 1562 bytes --]

On Fri, 4 Feb 2022, Eli Zaretskii wrote:

                     Please step through the code of insert-directory
> and describe here what you see there when it reports the error.
> Everything else is just noise, as far as investigation of this problem
> is concerned.

Using Edebug on the instrumented insert-directory function narrows in
on lisp/files.el
'''
  7502                     (apply 'call-process
  7503                            insert-directory-program nil t nil
  7504                            (append
  7505                             (if (listp switches) switches
  7506                               (unless (equal switches "")
  7507                                 ;; Split the switches at any spaces so we can
  7508                                 ;; pass separate options as separate args .
  7509                                 (split-string-and-unquote switches)))
  7510                             ;; Avoid lossage if FILE starts with `-'.
  7511                             '("--")
  7512                             (list file))))))
'''

The "result" exit status is "1" on line 7512 closing parenthesises.

The call is "ls -al -- /usr/X" and the problem is Plan's ls doesn't
take the "-a" switch.  For example

'''
% ls -a
usage: ls [-dlmnpqrstuFQ] [file ...]
% echo $?
1
% ls -al
usage: ls [-dlmnpqrstuFQ] [file ...]
% echo $?
1
'''

The Edebug steps continue to line 7581.

lisp/files.el:7581
"Listing directory failed but `access-file' worked"

In the case of Plan9's ls what is needed is to leave out the "-a" switch.

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Sat, 5 Feb 2022 03:07:07 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: luangruo@yahoo.com, 53739@debbugs.gnu.org
> 
> The "result" exit status is "1" on line 7512 closing parenthesises.
> 
> The call is "ls -al -- /usr/X" and the problem is Plan's ls doesn't
> take the "-a" switch.  For example
> 
> '''
> % ls -a
> usage: ls [-dlmnpqrstuFQ] [file ...]
> % echo $?
> 1
> % ls -al
> usage: ls [-dlmnpqrstuFQ] [file ...]
> % echo $?
> 1
> '''

Thanks, the problem is clear now.

> The Edebug steps continue to line 7581.
> 
> lisp/files.el:7581
> "Listing directory failed but `access-file' worked"
> 
> In the case of Plan9's ls what is needed is to leave out the "-a" switch.

Which means "C-u C-x d" should fix your use case nicely.  You can also
customize dired-listing-switches in your init file, if you intend to
use this version of 'ls' routinely.  It's a defcustom for this very
reason.

Bottom line: I think we can close this bug with no further action.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Sat, 5 Feb 2022, Eli Zaretskii wrote:

>> In the case of Plan9's ls what is needed is to leave out the "-a" switch.
>
> Which means "C-u C-x d" should fix your use case nicely.  You can also
> customize dired-listing-switches in your init file, if you intend to
> use this version of 'ls' routinely.  It's a defcustom for this very
> reason.
>
> Bottom line: I think we can close this bug with no further action.
>

Thanks.

Another fix is to re-order Emacs's exec-path and put plan9/bin second 
last at the bottom.  Emacs could DWIM to the exec-path on seeing 
plan9/bin at the top.

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Sat, 5 Feb 2022 09:04:29 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: luangruo@yahoo.com, 53739@debbugs.gnu.org
> 
> On Sat, 5 Feb 2022, Eli Zaretskii wrote:
> 
> >> In the case of Plan9's ls what is needed is to leave out the "-a" switch.
> >
> > Which means "C-u C-x d" should fix your use case nicely.  You can also
> > customize dired-listing-switches in your init file, if you intend to
> > use this version of 'ls' routinely.  It's a defcustom for this very
> > reason.
> >
> > Bottom line: I think we can close this bug with no further action.
> >
> 
> Thanks.
> 
> Another fix is to re-order Emacs's exec-path and put plan9/bin second 
> last at the bottom.  Emacs could DWIM to the exec-path on seeing 
> plan9/bin at the top.

I see no need to change Emacs for such a corner use case that already
has an easy solution.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Sat, 5 Feb 2022, Eli Zaretskii wrote:

>
> I see no need to change Emacs for such a corner use case that already
> has an easy solution.
>

Anyway.  I wonder what the Lisp Machine designers would have done for 
this corner case in the sense of furniture design and comfort.  For 
me, Emacs feels like furniture at times.  Edebug was an easy to use 
experience and well designed.

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> Bottom line: I think we can close this bug with no further action.

I agree.  Now done.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Fri, 4 Feb 2022 08:59:11 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: Po Lu <luangruo@yahoo.com>, 53739@debbugs.gnu.org
> 
> On Fri, 4 Feb 2022, Eli Zaretskii wrote:
> 
> >
> > The problem, as I explained up-thread, seems to be that 'ls' returns a
> > non-zero exit status, which insert-directory takes as a sign that the
> > directory listing didn't work.
> 
> The "9 man intro" page has this paragraph containing the "exit" keyword.
> 
> '''
>       DIAGNOSTICS
>            In Plan 9, a program's exit status is an arbitrary text
>            string, while on Unix it is an integer.  Section (1) of this
>            manual describes commands as though they exit with string
>            statuses.  In fact, exiting with an empty status corresponds
>            to exiting with status 0, and exiting with any non-empty
>            string corresponds to exiting with status 1.  See exits(3).
> '''
> 
> >
> > So: what does "ls -l" return as exit status when using the plan9
> > variant of 'ls'?
> >
> 
> Starting 'emacs -Q' at the end of this typescript reproduces the problem.
> 
> '''
> ; on GNU/Linux
> % echo $SHELL
> /usr/bin/zsh
> % PATH=/usr/local/plan9/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin; export PATH
> % echo $PATH
> /usr/van/.opam/cs3110-2-21fa/bin:/usr/local/plan9/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin
> % which ls
> /usr/local/plan9/bin/ls
> % ls -l
> --rw-r--r-- M 0 van dialout 46 Feb  4 18:22 emacs28
> --rw-r--r-- M 0 van dialout 47 Feb  4 18:14 emacs28q
> % echo $?
> 0
> % /usr/local/plan9/bin/ls -l
> --rw-r--r-- M 0 van dialout 46 Feb  4 18:22 emacs28
> --rw-r--r-- M 0 van dialout 47 Feb  4 18:14 emacs28q
> % echo $?
> 0

Thanks, but we need to know what Emacs sees as the exit status inside
insert-directory.  So, as I already said, the best approach top
investigating this is to step with Edebug through the code in
insert-directory, and see what kind of exit status we receive there,
and why we decide that "directory listing failed".

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

On Thu, 3 Feb 2022, Po Lu wrote:

>
> It would be a great deal of trouble to support their version of ls (and
> indeed, their versions of many other common Unix utilities such as `rm'
> and `cp'), so I think the answer is simply "don't run Emacs with
> plan9/bin at the head of your PATH".
>

Maybe the user can have their cake and eat it.  What if Emacs ignores 
the PATH element containing plan9/bin and the shell environment 
outside Emacs is allowed to prefer plan9's commands?

-- 
vl

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Eli Zaretskii]

> Date: Thu, 3 Feb 2022 13:05:27 +0000 (UTC)
> From: Van Ly <van.ly@sdf.org>
> cc: Eli Zaretskii <eliz@gnu.org>, 53739@debbugs.gnu.org
> 
> Maybe the user can have their cake and eat it.  What if Emacs ignores 
> the PATH element containing plan9/bin and the shell environment 
> outside Emacs is allowed to prefer plan9's commands?

Emacs doesn't use PATH, it uses exec-path.

bug#53739: 28.0.91; "C-x d" fails when "emacs -Q" starts on NetBSD with plan9/bin on head of PATH

[Van Ly]

[-- Attachment #1: Type: text/plain, Size: 1940 bytes --]

On Thu, 3 Feb 2022, Eli Zaretskii wrote:

>
> I guess the version of 'ls' there is incompatible or something?  The
> message means that 'ls' exited with a non-zero status for some reason.
> Please investigate more what went wrong there and report here the
> details, so we could decide whether we want to support "plan9/bin"
> idiosyncrasies, whatever they are.  Stepping with Edebug through
> insert-directory should help  you understand what's going on.
>

I guess plan9/bin contains many commands identical in name to those 
GNU Emacs expects from POSIX but they have plan9's reinvention.

One fix is not to put plan9/bin on the head of PATH but at the tail 
and GNU Emacs will succeed at "C-x d" or dired.  Maybe a response 
like how "dired-use-ls-dired" is handled will point the user in the 
right direction.

> Also, please explain what you mean by "for the following interactive
> shells".

The interactive shell is either the shell given at login by setting 
as follows then logout, login

  > usermod -s /path/to/shell username

and that shell is listed for the username in  /etc/passwd

or I "exec /path/to/shell" to change the shell; I think
shell-file-name is set to the value listed in /etc/passwd.

I had the shells sh, zsh and rc running interactively at the command prompt.

>            How does the interactive shell enter this picture?

The interactive shell is the environment on the command line where I 
launch "emacs -Q".

                                                                And what
> is the value of shell-file-name in the session where "C-x d" fails?

Two examples of the value for shell-file-name in the session where 
"C-x d" fails are as follows:
'''
;; set shell to sh by usermod, logout and login
Its value is "/bin/sh"
Original value was "/usr/pkg/bin/zsh"

;; set shell to rc by usermod, logout and login
Its value is "/usr/local/plan9/bin/rc"
Original value was "/usr/pkg/bin/zsh"
'''

-- 
vl