This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Building perl-5.10.0

On Apr  1 18:38, Eric Blake wrote:
> According to Corinna Vinschen on 4/1/2008 7:44 AM:
> |  Shouldn't the "nobody" entry
> | disappear when calling chmod?  That's how I understand the statement in
> | the POSIX docs:
> |
> |  "An alternate file access control mechanism shall [...] be disabled for
> |   a file after the file permission bits are changed for that file with
> |   chmod( ).  The disabling of the alternate mechanism need not disable
> |   any additional mechanisms supported by an implementation."
> |
> | Either the ACLs of a file are not an "alternate" access mechanism,
> | but an "additional" access mechanism.  But that doesn't match the
> | description either:
> |
> |  "An additional access control mechanism shall only further restrict
> |   the access permissions defined by the file permission bits."
> Yes, those were the two paragraphs I was noticing when I made my claim
> that cygwin's chmod(2) wasn't obeying POSIX.
> ACLs can serve as both "alternate" (give more rights to some users than
> what is implied by the traditional stat bits) and "additional" (restrict
> rights to certain users outside of what is shown in the traditional stat
> bits).
> | Or, Linux doesn't follow POSIX here, which seems unlikely to me.
> Actually, it seems highly likely to me - after all, at one point, POSIX
> considered standardizing a form of ACLs, but it never went anywhere (and
> in the meantime, several competing styles of how to implement ACLs cropped
> up; Solaris and Linux tackle the issue noticably different, and Selinux
> security descriptors are yet another wrinkle in the picture).

In the meantime I tested this scenario on Solaris 9 as well and I found
that it behaves exactly as Linux and Cygwin.  It's nice to see that
both, Linux and Solaris, are following Cygwin's lead here <cough, cough>.
Just for the records (again), Cygwin's ACL support is modeled on the
Solaris ACL API.

>   Maybe it's
> worth asking on the Austin Group mailing list?

I think so, yes.  It looks somewhat unusual if two important OSes
seem to contradict what's in the specs.  Are you going to ask?


Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat

Unsubscribe info:
Problem reports:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]