Re: Does "nobody" is considered as group-member of others?<eca073b1-308a-436e-9f09-fa0d4c127eb7@j28g2000hsj.googlegroups.com>

On 2008-02-28, Fro wrote:
>
> the operating system (Unix) considers a php-server as a user with name
> "nobody". For example, if my php-script saves a file uploaded by a
> user, the owner of the file will be "nobody". I would like to know if
> "nobody" is considered as "group" or "others" (in terms of the "chmod"
> command)? In other words, is "chmod g+w dirname" sufficient to allow
> to "nobody" to write in the directory "dirname" or should I use "chmod
> o+w dirname"?

If 'nobody' is a member of the group that owns "dirname", then
group write access is enough, otherwise it needs world (i.e.,
other) write permissions.


--
Chris F.A. Johnson, author | <http://cfaj.freeshell.org>
Shell Scripting Recipes: | My code in this post, if any,
A Problem-Solution Approach | is released under the
2005, Apress | GNU General Public Licence

On Feb 27, 9:24 pm, "Chris F.A. Johnson" <cfajohn...@gmail.com> wrote:
> On 2008-02-28, Fro wrote:
>
> > the operating system (Unix) considers a php-server as a user with name
> > "nobody". For example, if my php-script saves a file uploaded by a
> > user, the owner of the file will be "nobody". I would like to know if
> > "nobody" is considered as "group" or "others" (in terms of the "chmod"
> > command)? In other words, is "chmod g+w dirname" sufficient to allow
> > to "nobody" to write in the directory "dirname" or should I use "chmod
> > o+w dirname"?
>
> If 'nobody' is a member of the group that owns "dirname", then
> group write access is enough, otherwise it needs world (i.e.,
> other) write permissions.
>
But this is the question. Is nobody a group member?

What is that???
<eca073b1-308a-436e-9f09-fa0d4c127eb7@j28g2000hsj.googlegroups.com>

On 2008-02-28, Fro wrote:
> On Feb 27, 9:24 pm, "Chris F.A. Johnson" <cfajohn...@gmail.com> wrote:
>> On 2008-02-28, Fro wrote:
>>
>> > the operating system (Unix) considers a php-server as a user with name
>> > "nobody". For example, if my php-script saves a file uploaded by a
>> > user, the owner of the file will be "nobody". I would like to know if
>> > "nobody" is considered as "group" or "others" (in terms of the "chmod"
>> > command)? In other words, is "chmod g+w dirname" sufficient to allow
>> > to "nobody" to write in the directory "dirname" or should I use "chmod
>> > o+w dirname"?
>>
>> If 'nobody' is a member of the group that owns "dirname", then
>> group write access is enough, otherwise it needs world (i.e.,
>> other) write permissions.
>>
> But this is the question. Is nobody a group member?

There are many groups. What group owns the directory?

What groups does 'nobody' belong to?

These answers are system specific; your *nix system could be
different from mine.

The commonly used group for 'nobody', in my experience, is 65534,
and is called either 'nobody' or 'nogroup'. On another system, it
is group 550, on yet another, it is 99, both called 'nobody'.

Unless the group ownership of the directory is the same as nobody's
group, you will need world write permissions.

Under normal circumstances, 'nobody' only belongs to the one group,
and no other users belong to it. The sysadmin can change that.


--
Chris F.A. Johnson, author | <http://cfaj.freeshell.org>
Shell Scripting Recipes: | My code in this post, if any,
A Problem-Solution Approach | is released under the
2005, Apress | GNU General Public Licence

On Feb 27, 9:58 pm, "Chris F.A. Johnson" <cfajohn...@gmail.com> wrote:
> On 2008-02-28, Fro wrote:
> > On Feb 27, 9:24 pm, "Chris F.A. Johnson" <cfajohn...@gmail.com> wrote:
> >> On 2008-02-28, Fro wrote:
>
> >> > the operating system (Unix) considers a php-server as a user with name
> >> > "nobody". For example, if my php-script saves a file uploaded by a
> >> > user, the owner of the file will be "nobody". I would like to know if
> >> > "nobody" is considered as "group" or "others" (in terms of the "chmod"
> >> > command)? In other words, is "chmod g+w dirname" sufficient to allow
> >> > to "nobody" to write in the directory "dirname" or should I use "chmod
> >> > o+w dirname"?
>
> >> If 'nobody' is a member of the group that owns "dirname", then
> >> group write access is enough, otherwise it needs world (i.e.,
> >> other) write permissions.
>
> > But this is the question. Is nobody a group member?
>
> There are many groups. What group owns the directory?
>
> What groups does 'nobody' belong to?
>
> These answers are system specific; your *nix system could be
> different from mine.
>
> The commonly used group for 'nobody', in my experience, is 65534,
> and is called either 'nobody' or 'nogroup'. On another system, it
> is group 550, on yet another, it is 99, both called 'nobody'.
>
> Unless the group ownership of the directory is the same as nobody's
> group, you will need world write permissions.
>
> Under normal circumstances, 'nobody' only belongs to the one group,
> and no other users belong to it. The sysadmin can change that.
>
> --
> Chris F.A. Johnson, author | <http://cfaj.freeshell.org>
> Shell Scripting Recipes: | My code in this post, if any,
> A Problem-Solution Approach | is released under the
> 2005, Apress | GNU General Public Licence

OK. I think I got it. It type:
groups my_user_name

Then I type:
groups nobody

In this way I see that there are no overlap between the two sets of
groups. So, I need to use "chmod o+r" to allow "nobody" to read my
files and directories.

On Thu, 28 Feb 2008 03:33:10 +0100, Fro <showandbeshown@gmail.com> wrote:

> What is that???
> <eca073b1-308a-436e-9f09-fa0d4c127eb7@j28g2000hsj.googlegroups.com>

An error in his newscclient (slrn/0.9.8.1pl2 ), which drops the
'References:' text.

From comp.infosystems.www.authoring.html:

On Wed, 27 Feb 2008 00:50:58 +0100, Blinky the Shark <no.spam@box.invalid>
wrote:
> andrew wrote:
>> The latest svn slrn is very, very stable and might be a better choice
>> than slrn/0.9.8.1pl2. Some discussion in news.software.readers about
>> the lost 'References' header if you are interested.
>
> And from that group, where I took this issue earlier today, this is from
> the author of that version:
>
> <q>
> 0.9.8.1pl2 is buggy in its treatment of headers with long "words". As
> such, anything can happen. In this case, version 0.9.8.1pl2 wrapped
> the references header but because of a bug, the line containing
> "References:" was deleted. As a result, the remaining continuation
> lines became associated with the previous header: the subject header,
> </q>


--
Rik Wasmus