bug#819: 23.0.60; group and owner "Everyone" - what's that about?

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

emacs -Q
Dired on a subdir of, say, c:\
 
.. has the owner and group listed as "Everyone". What's that about? If
I click `..', I see that none of the files or subdirectories in c:\
belong to "Everyone" (in my case). That "Everyone" artifact is
incorrect and misleading.
 

In GNU Emacs 23.0.60.1 (i386-mingw-nt5.1.2600)
 of 2008-08-01 on LENNART-69DE564
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4) --no-opt --cflags -Ic:/g/include
-fno-crossjumping'
 

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Date: Fri, 29 Aug 2008 08:00:39 -0700
> Cc: 
> 
> emacs -Q
> Dired on a subdir of, say, c:\
>  
> .. has the owner and group listed as "Everyone". What's that about? If
> I click `..', I see that none of the files or subdirectories in c:\
> belong to "Everyone" (in my case). That "Everyone" artifact is
> incorrect and misleading.

You interpret "Everyone" too literally.  "Everyone" is the Windows way
of attributing a file's ownership to no one in particular.  This is
what you will see for each and every file on a FAT32 volume, because
FAT32 does not support file security, only NTFS does.

So "Everyone" is not an artefact, it's what Windows reports to Emacs
as the owner and primary group of a file or directory.

That said, I cannot reproduce this on my machine, with NTFS
filesystem: each file and directory I see belongs to some valid user
or group, and in particular all subdirectories of c:\ have
"Administrators" as the owner of ".." and "SYSTEM" as its primary
group.  That is what I'd expect on a typical Windows machine with NTFS
filesystem.

So what you see is some strange peculiarity of your system, I think.
Perhaps because you upgraded your filesystem from FAT32?

In any case, "Everyone" is a legitimate value of a Windows owner
and/or primary group of a file.

bug#819: marked as done (23.0.60; group and owner "Everyone" - what's that about?)

[Emacs bug Tracking System]

[-- Attachment #1: Type: text/plain, Size: 871 bytes --]


Your message dated Fri, 29 Aug 2008 21:01:12 +0300
with message-id <u8wufaefr.fsf@gnu.org>
and subject line bug#819: 23.0.60; group and owner "Everyone" - what's that about?
has caused the Emacs bug report #819,
regarding 23.0.60; group and owner "Everyone" - what's that about?
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact don@donarmstrong.com
immediately.)


-- 
819: http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=819
Emacs Bug Tracking System
Contact don@donarmstrong.com with problems

[-- Attachment #2: Type: message/rfc822, Size: 3030 bytes --]

From: "Drew Adams" <drew.adams@oracle.com>
To: <emacs-pretest-bug@gnu.org>
Subject: 23.0.60; group and owner "Everyone" - what's that about?
Date: Fri, 29 Aug 2008 08:00:39 -0700
Message-ID: <001201c909e8$001b46a0$0200a8c0@us.oracle.com>

emacs -Q
Dired on a subdir of, say, c:\
 
.. has the owner and group listed as "Everyone". What's that about? If
I click `..', I see that none of the files or subdirectories in c:\
belong to "Everyone" (in my case). That "Everyone" artifact is
incorrect and misleading.
 

In GNU Emacs 23.0.60.1 (i386-mingw-nt5.1.2600)
 of 2008-08-01 on LENNART-69DE564
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4) --no-opt --cflags -Ic:/g/include
-fno-crossjumping'
 




[-- Attachment #3: Type: message/rfc822, Size: 1283 bytes --]

From: Eli Zaretskii <eliz@gnu.org>
To: 819-done@emacsbugs.donarmstrong.com
Subject: bug#819: 23.0.60; group and owner "Everyone" - what's that about?
Date: Fri, 29 Aug 2008 21:01:12 +0300
Message-ID: <u8wufaefr.fsf@gnu.org>

This is not a bug, but expected behavior, for files/directories for
which NTFS file security records specify "Everyone" (numerical value
of zero) for the file's owner or primary group.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

> > emacs -Q
> > Dired on a subdir of, say, c:\
> >  
> > .. has the owner and group listed as "Everyone". What's 
> > that about? If I click `..', I see that none of the files or
> > subdirectories in c:\ belong to "Everyone" (in my case). That
> > "Everyone" artifact is incorrect and misleading.
> 
> You interpret "Everyone" too literally.  "Everyone" is the Windows way
> of attributing a file's ownership to no one in particular.  This is
> what you will see for each and every file on a FAT32 volume, because
> FAT32 does not support file security, only NTFS does.
> 
> So "Everyone" is not an artefact, it's what Windows reports to Emacs
> as the owner and primary group of a file or directory.

In Emacs 23, it's not what I see in Dired for the other files and directories,
besides `..'. In my case, I see my user name as the owner for each entry except
`..', and I see `None' as the group for each entry except `..'.

In Emacs 22, I see my user name as the owner for each entry (including `..') and
`root' as the group for each entry (including `..'). In Emacs 20, I see my user
name as the owner for each entry (including `..') and `5' as the group for each
entry (including `..').

IOW, before Emacs 23, (1) `..' was not treated differently from other entries
and (2) the group was not `None'. Change #2 seems OK to me, but I don't
understand #1: Why should `..' be different? 

And why should following the link `..' show something different from what was
shown for `..' in the subdir? The properties of a given directory should always
appear the same, regardless of where you are viewing them.

> That said, I cannot reproduce this on my machine, with NTFS
> filesystem: each file and directory I see belongs to some valid user
> or group, and in particular all subdirectories of c:\ have
> "Administrators" as the owner of ".." and "SYSTEM" as its primary
> group.  That is what I'd expect on a typical Windows machine with NTFS
> filesystem.

My hard disk is formatted FAT32, not NTFS.

> So what you see is some strange peculiarity of your system, I think.
> Perhaps because you upgraded your filesystem from FAT32?

No. It was and is still FAT32.

> In any case, "Everyone" is a legitimate value of a Windows owner
> and/or primary group of a file.

Whatever is used should be consistent: (1) `..' shouldn't be special, and (2)
`..' should show the same properties (owner group) as what is shown for it when
you follow the `..' link - the directory's properties should not change
depending on where you are looking at them.

bug#819: closed by Eli Zaretskii <eliz@gnu.org> (bug#819: 23.0.60; group and owner "Everyone" - what's that about?)

[Drew Adams]

> It has been closed by Eli Zaretskii <eliz@gnu.org>.
> Their explanation is attached below along with your original report.
> If this explanation is unsatisfactory and you have not received a
> better one in a separate message then please contact Eli 
> Zaretskii <eliz@gnu.org> by replying to this email.

I don't agree. You closed this even before my reply to your mail, and I replied
immediately. You supposed that I had NTFS, which I do not.

It is a bug that `..' is treated differently from other Dired entries, and only
when it represents the top-level directory (e.g. c:\).

It is only when `..' stands for the top-level directory that `Everyone Everyone'
appears. In all other cases, all Dired entries, including `..' entries, list the
owner as my user name and the group as `None'.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Jason Rumney]

Drew Adams wrote:
> IOW, before Emacs 23, (1) `..' was not treated differently from other entries
> and (2) the group was not `None'. Change #2 seems OK to me, but I don't
> understand #1: Why should `..' be different? 
>   

Prior to Emacs 23, the Windows port just made up ownership on the spot, 
to avoid the bug reports about unexpected results trying to map from 
NTFS Access Control Lists onto traditional POSIX ownership.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Cc: <emacs-pretest-bug@gnu.org>, <bug-gnu-emacs@gnu.org>
> Date: Fri, 29 Aug 2008 11:12:35 -0700
> 
> In Emacs 23, it's not what I see in Dired for the other files and directories,
> besides `..'. In my case, I see my user name as the owner for each entry except
> `..', and I see `None' as the group for each entry except `..'.

"None" is the name of the primary group on a stand-alone workstation
(probably because someone at MS thought that groups only make sense
when there's a domain server).  If your machine is a stand-alone
workstation, "None" is what I'd expect.

> In Emacs 22, I see my user name as the owner for each entry (including `..') and
> `root' as the group for each entry (including `..'). In Emacs 20, I see my user
> name as the owner for each entry (including `..') and `5' as the group for each
> entry (including `..').

This code evolved since it was written.  Emacs 20 to 22 didn't
actually ask the filesystem for the owner and primary group of each
file.  Instead, they retrieved during startup the name of the user who
runs Emacs, and then attributed each file to that user, and just
invented its group as "root".  (`5' in Emacs 20 is just due to a bug
in understanding the components of the Windows Security ID
descriptor.)  In Emacs 23, we ask the filesystem about the owner and
the group, and report that if the information is available.

This is why each version behaves differently.  But it is important to
know that before Emacs 23, the displayed info was largely invented
rather than extracted from the filesystem.  Now you see what is really
there (if you are on NTFS, which is not your case, see below).

> And why should following the link `..' show something different from what was
> shown for `..' in the subdir?

You mean the link `.' (on dot) should show the same as what was shown
for `..' (2 dots) in a subdir, right?  If that's not happening for
you, please show a concrete example.

> My hard disk is formatted FAT32, not NTFS.

On a FAT32 we currently attribute all files and directories to the
user who runs Emacs.  FAT32 fails the file-security syscalls, so we
fall back on what Emacs 22 did, because files that have no owner are
in a way ``owned'' by everyone, the current user included.

> Whatever is used should be consistent: (1) `..' shouldn't be special

`..' isn't treated special in the Emacs code that deals with file
ownership.

> and (2)
> `..' should show the same properties (owner group) as what is shown for it when
> you follow the `..' link - the directory's properties should not change
> depending on where you are looking at them.

If you follow the `..' link twice, you mean, right?  Or if you follow
it once, then look at the `.' line.  Right?

bug#819: closed by Eli Zaretskii <eliz@gnu.org> (bug#819: 23.0.60; group and owner "Everyone" - what's that about?)

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Date: Fri, 29 Aug 2008 11:44:07 -0700
> Cc: 
> 
> > It has been closed by Eli Zaretskii <eliz@gnu.org>.
> > Their explanation is attached below along with your original report.
> > If this explanation is unsatisfactory and you have not received a
> > better one in a separate message then please contact Eli 
> > Zaretskii <eliz@gnu.org> by replying to this email.
> 
> I don't agree. You closed this even before my reply to your mail, and I replied
> immediately.

I closed it because it complained about Emacs showing "Everyone" as a
group name.  "Everyone" is a legitimate Windows group name, so the bug
as submitted is actually expected behavior on Windows.

If you want to complain about something else, please submit another
bug report, and please change the description of a bug; telling that
"Everyone" is some kid of artefact is simply incorrect, and I can do
nothing but close such bugs.

> You supposed that I had NTFS, which I do not.

No, I supposed you had FAT32.  I don't expect "Everyone" anywhere
else.  I just didn't explain myself clearly.

> It is a bug that `..' is treated differently from other Dired entries, and only
> when it represents the top-level directory (e.g. c:\).

That's not what the code does.  There's no special case for `..'.

> It is only when `..' stands for the top-level directory that `Everyone Everyone'
> appears. In all other cases, all Dired entries, including `..' entries, list the
> owner as my user name and the group as `None'.

I don't see that on FAT32 volume to which I have access.

The code that reports user and group IDs uses fallbacks for when the
file security syscalls fail.  The fallback is to attribute the files
to the user who runs Emacs.  On FAT32 volumes the syscalls fail, so
what you see is our best shot on reporting something sensible.

bug#819: closed by Eli Zaretskii <eliz@gnu.org> (bug#819: 23.0.60; group and owner "Everyone" - what's that about?)

[Eli Zaretskii]

> Date: Fri, 29 Aug 2008 22:38:06 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: bug-gnu-emacs@gnu.org, bug-submit-list@donarmstrong.com,
> 	819@emacsbugs.donarmstrong.com
> 
> The code that reports user and group IDs uses fallbacks for when the
> file security syscalls fail.  The fallback is to attribute the files
> to the user who runs Emacs.  On FAT32 volumes the syscalls fail, so
> what you see is our best shot on reporting something sensible.

Actually, I take that back: these syscalls should NOT fail on FAT32.
You should see "Everyone" for each file and directory on FAT32.  The
fact that you see your username means that something causes the
file-security syscalls to fail.  Perhaps that's the same problem as
the one reported by Juanma a few days ago (on NTFS).  We will see when
I debug that one.

> I don't see that on FAT32 volume to which I have access.

This part is still true.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

> > And why should following the link `..' show something 
> > different from what was
> > shown for `..' in the subdir?
> 
> You mean the link `.' (on dot) should show the same as what was shown
> for `..' (2 dots) in a subdir, right?  If that's not happening for
> you, please show a concrete example.

I probably misspoke a bit. The `Everyone'...`Everyone' owner and group appear
only on `..', and only when `..' represents the top level: c:\. And for Dired in
c:\ itself there is no `.' entry, so I can't say it contradicts what `..' showed
one level down (`Everyone'...`Everyone').

The bug, as I see it, is that `..' is different (showing
`Everyone'...`Everyone') _only_ for a directory just under c:\ - for example,
Dired in c:\foo\. Everywhere else, I see the same thing for `..' that I see for
every other file or directory (my user name and `None', in my case).

> > My hard disk is formatted FAT32, not NTFS.
> 
> On a FAT32 we currently attribute all files and directories to the
> user who runs Emacs.  FAT32 fails the file-security syscalls, so we
> fall back on what Emacs 22 did, because files that have no owner are
> in a way ``owned'' by everyone, the current user included.

I see. But I do not in fact see the same thing in Emacs 22 as in Emacs 23, so
you must be doing something different in this case.

bug#819: closed by Eli Zaretskii <eliz@gnu.org> (bug#819: 23.0.60; group and owner "Everyone" - what's that about?)

[Drew Adams]

[-- Attachment #1: Type: text/plain, Size: 1832 bytes --]

> I closed it because it complained about Emacs showing "Everyone" as a
> group name.  "Everyone" is a legitimate Windows group name, so the bug
> as submitted is actually expected behavior on Windows.

What I reported (or meant to report) was about the top level (c:\) being treated
different from other levels in Dired: the fact that `..' in a directory just
under the top level doesn't show the same properties as elsewhere.

I might not have made that clear enough, but this is the problem I see: treating
`..' differently in this one case.

> If you want to complain about something else, please submit another
> bug report, and please change the description of a bug; telling that
> "Everyone" is some kid of artefact is simply incorrect, and I can do
> nothing but close such bugs.

Bug #819 and its subject line are adequate for this bug, IMO - no need to file
another bug. I've clarified what I see. If you want to fix it, OK; if not, OK.

> > You supposed that I had NTFS, which I do not.
> 
> No, I supposed you had FAT32.  I don't expect "Everyone" anywhere
> else.  I just didn't explain myself clearly.

I too didn't explain myself clearly enough. Hopefully, it's clear now.

> > It is a bug that `..' is treated differently from other 
> > Dired entries, and only
> > when it represents the top-level directory (e.g. c:\).
> 
> That's not what the code does.  There's no special case for `..'.

I hope my description is clear now. I'm looking at the Dired entries, not at the
code.

> > It is only when `..' stands for the top-level directory 
> > that `Everyone Everyone' appears. In all other cases, all Dired
> > entries, including `..' entries, list the
> > owner as my user name and the group as `None'.
> 
> I don't see that on FAT32 volume to which I have access.

Attached is a screenshot, in case it helps.


[-- Attachment #2: bug-dired-owner-group.png --]
[-- Type: image/png, Size: 17623 bytes --]

bug#819: closed by Eli Zaretskii <eliz@gnu.org> (bug#819: 23.0.60; group and owner "Everyone" - what's that about?)

[Drew Adams]

> > The code that reports user and group IDs uses fallbacks for when the
> > file security syscalls fail.  The fallback is to attribute the files
> > to the user who runs Emacs.  On FAT32 volumes the syscalls fail, so
> > what you see is our best shot on reporting something sensible.
> 
> Actually, I take that back: these syscalls should NOT fail on FAT32.
> You should see "Everyone" for each file and directory on FAT32.  The
> fact that you see your username means that something causes the
> file-security syscalls to fail.  Perhaps that's the same problem as
> the one reported by Juanma a few days ago (on NTFS).  We will see when
> I debug that one.

If it always showed Everyone, that would be OK. The problem is that it is not
consistent (on my machine, at least), treating directories just under the top
level differently from others.

> > I don't see that on FAT32 volume to which I have access.
> 
> This part is still true.

See the screenshot I sent. It sounds like we are seeing different things. I
definitely have FAT32, FWIW.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Stefan Monnier]

> The bug, as I see it, is that `..' is different (showing
> `Everyone'...`Everyone') _only_ for a directory just under c:\ - for
> example, Dired in c:\foo\. Everywhere else, I see the same thing for
> `..' that I see for every other file or directory (my user name and
> `None', in my case).

What Eli says is that this "bug" is not in Emacs.  Emacs just reports
what the Windows API reports in this case.


        Stefan

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Jason Rumney]

Drew Adams wrote:

> I probably misspoke a bit. The `Everyone'...`Everyone' owner and group appear
> only on `..', and only when `..' represents the top level: c:\. And for Dired in
> c:\ itself there is no `.' entry, so I can't say it contradicts what `..' showed
> one level down (`Everyone'...`Everyone').
>
> The bug, as I see it, is that `..' is different (showing
> `Everyone'...`Everyone') _only_ for a directory just under c:\ - for example,
> Dired in c:\foo\. Everywhere else, I see the same thing for `..' that I see for
> every other file or directory (my user name and `None', in my case).
>   

So what seems to be happening is that the security descriptor for the 
disk itself is different than the security descriptor for the files and 
folders it contains, and Emacs is reporting that.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

> > I probably misspoke a bit. The `Everyone'...`Everyone' 
> owner and group appear
> > only on `..', and only when `..' represents the top level: 
> c:\. And for Dired in
> > c:\ itself there is no `.' entry, so I can't say it 
> contradicts what `..' showed
> > one level down (`Everyone'...`Everyone').
> >
> > The bug, as I see it, is that `..' is different (showing
> > `Everyone'...`Everyone') _only_ for a directory just under 
> c:\ - for example,
> > Dired in c:\foo\. Everywhere else, I see the same thing for 
> `..' that I see for
> > every other file or directory (my user name and `None', in my case).
> 
> So what seems to be happening is that the security descriptor for the 
> disk itself is different than the security descriptor for the 
> files and folders it contains, and Emacs is reporting that.

Just so I can understand a little better, could you elaborate? I thought that
Eli was saying that with FAT32 it should say Everyone...Everyone everywhere.
Which is more or less (or differently?) secure, my disk or the files and
folders? How is that determined/defined in the Windows UI? Thx.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Jason Rumney]

Drew Adams wrote:
> Just so I can understand a little better, could you elaborate? I thought that
> Eli was saying that with FAT32 it should say Everyone...Everyone everywhere.
>   

That is what I see with a FAT32 disk too. I have no idea why you see 
different, but it probably has something to do with your local Windows 
configuration.

> Which is more or less (or differently?) secure, my disk or the files and
> folders? How is that determined/defined in the Windows UI? Thx.
>   

You are asking questions that only someone with access to Windows source 
code could answer. It may not be different security, the security calls 
may be failing, causing Emacs to fall back on the old strategy of making 
something up.

>
>   

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

> > Just so I can understand a little better, could you 
> > elaborate? I thought that Eli was saying that with FAT32
> > it should say Everyone...Everyone everywhere.
> 
> That is what I see with a FAT32 disk too. I have no idea why you see 
> different, but it probably has something to do with your 
> local Windows configuration.
> 
> > Which is more or less (or differently?) secure, my disk or 
> > the files and folders? How is that determined/defined in the
> > Windows UI? Thx.
> >   
> 
> You are asking questions that only someone with access to 
> Windows source code could answer. It may not be different
> security, the security calls may be failing, causing Emacs
> to fall back on the old strategy of making something up.

OK, I guess the bottom line is that my Windows installation/config is in some
way, for some reason, exceptional, and this is not an Emacs bug. Thanks for
looking into it; sorry for the noise.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Cc: "'Eli Zaretskii'" <eliz@gnu.org>
> Date: Fri, 29 Aug 2008 19:00:12 -0700
> 
> > > I probably misspoke a bit. The `Everyone'...`Everyone' 
> > owner and group appear
> > > only on `..', and only when `..' represents the top level: 
> > c:\. And for Dired in
> > > c:\ itself there is no `.' entry, so I can't say it 
> > contradicts what `..' showed
> > > one level down (`Everyone'...`Everyone').
> > >
> > > The bug, as I see it, is that `..' is different (showing
> > > `Everyone'...`Everyone') _only_ for a directory just under 
> > c:\ - for example,
> > > Dired in c:\foo\. Everywhere else, I see the same thing for 
> > `..' that I see for
> > > every other file or directory (my user name and `None', in my case).
> > 
> > So what seems to be happening is that the security descriptor for the 
> > disk itself is different than the security descriptor for the 
> > files and folders it contains, and Emacs is reporting that.
> 
> Just so I can understand a little better, could you elaborate? I thought that
> Eli was saying that with FAT32 it should say Everyone...Everyone everywhere.

Yes, that is what I was saying.  The fact that some files are not
reported as Everyone...Everyone might be some bug in Emacs, I will
look into this shortly.

What Jason suggests above is AFAIK impossible, since Windows security
APIs always report the same descriptor S-1-1-0 for every file on a
FAT32 volume.  S-1-1-0 is the security descriptor of "Everyone".

Do you have the subinacl.exe program?  (If not, you can download and
install the Windows Resource Kit.)  Try this command and see what it
reports about your directories:

    subinacl /output=log.txt /subdirectories c:\some\directory

(This could take a while on a large disk, so pick up a directory that
does not have too many subdirectories and files under it.)  The
results are in the file log.txt that is encoded in UTF-16, and should
be viewed by visiting it with Emacs.

Do you see anything but "everyone" in the reported owner and primary
group of your files?

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Cc: <819@emacsbugs.donarmstrong.com>, "'Eli Zaretskii'" <eliz@gnu.org>
> Date: Fri, 29 Aug 2008 20:29:15 -0700
> 
> sorry for the noise.

Wait with that apology; I might find some Emacs bug after all.  Stay
tuned.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Eli Zaretskii]

> Date: Sat, 30 Aug 2008 11:15:09 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 819@emacsbugs.donarmstrong.com
> 
> > From: "Drew Adams" <drew.adams@oracle.com>
> > Cc: <819@emacsbugs.donarmstrong.com>, "'Eli Zaretskii'" <eliz@gnu.org>
> > Date: Fri, 29 Aug 2008 20:29:15 -0700
> > 
> > sorry for the noise.
> 
> Wait with that apology; I might find some Emacs bug after all.  Stay
> tuned.

OK, I think I know what is going on.  You are using an Emacs 23 binary
that was built before Aug 15 this year, right?

If so, there was a bug in the code back then which would manifest
itself when w32-get-true-file-attributes was at its default value
`local'.  The bug would cause Emacs to think that every file except
the root directory on each local drive was not local, and therefore
Emacs would not try to retrieve the file's security descriptor and
glean the owner and group from that.  Instead, Emacs would fall back
on attributing each file to the user who runs Emacs.

This is why you see the root directory owned by "Everyone", while the
rest of files are not.

To see if my theory is right, either install a binary built after Aug
15, or set w32-get-true-file-attributes to t and do a Dired on C:\.
If I'm right, you will see each and every file attributed to Everyone,
like I'd expect on a FAT32 volume.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

> OK, I think I know what is going on.  You are using an Emacs 23 binary
> that was built before Aug 15 this year, right?

Yes, August 1, 2008.

> If so, there was a bug in the code back then which would manifest
> itself when w32-get-true-file-attributes was at its default value
> `local'.  The bug would cause Emacs to think that every file except
> the root directory on each local drive was not local, and therefore
> Emacs would not try to retrieve the file's security descriptor and
> glean the owner and group from that.  Instead, Emacs would fall back
> on attributing each file to the user who runs Emacs.
> 
> This is why you see the root directory owned by "Everyone", while the
> rest of files are not.
> 
> To see if my theory is right, either install a binary built after Aug
> 15, or set w32-get-true-file-attributes to t and do a Dired on C:\.
> If I'm right, you will see each and every file attributed to Everyone,
> like I'd expect on a FAT32 volume.

Bravo! Gagne. That was it - setting w32-get-true-file-attributes to t shows
Everyone everywhere. Thanks for tracking this down. Sorry I didn't have a more
recent build; it would have saved a bug report and your debugging time.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Drew Adams]

FYI - Lennart uploaded a new binary (8-29), and it works (Everyone...Everyone,
everywhere) in that build. Thx to all who helped on this.

bug#819: 23.0.60; group and owner "Everyone" - what's that about?

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Cc: <bug-gnu-emacs@gnu.org>
> Date: Sat, 30 Aug 2008 06:55:15 -0700
> 
> Bravo! Gagne. That was it - setting w32-get-true-file-attributes to t shows
> Everyone everywhere. Thanks for tracking this down. Sorry I didn't have a more
> recent build; it would have saved a bug report and your debugging time.

No sweat: tracking this down made me realize that the Emacs 22.3
pretest has the same problem with local/remote drive detection, so I
fixed it in time for 22.3.

(Emacs 22.x does not display true file owner and group on local files,
but it still works harder to get file attributes such as the number of
hard links and ``inode'' value.)