RFC: [PATCH] Re: Test case for fakeroot-hurd failure with a socket

[Svante Signell]

On Thu, 2015-05-07 at 16:51 +0200, Samuel Thibault wrote:
> Svante Signell, le Thu 07 May 2015 10:14:23 +0200, a écrit :
> > I think I understand most of your writing above :)
> > 
> > Should the  netfs_set_translator stub be implemented in trans/fakeroot.c
> > or in libnetfs/set-get-trans.c?
> 
> The former.
> 
> > Still confusing is which functions are called: The ones in libnetfs vs
> > the ones in fakeroot?
> 
> The ones in libnetfs, and they call stubs from fakeroot.

Attached is a patch that solves the test case for sockets.

An attempt to explain:
libnetfs/file-set-translator.c:netfs_S_file_set_translator() hits the
default case in the newmode switch statement which in
netfs_attempt_chmod() returns an error of EOPNOTSUPP and hits the
fallback case calling the stub in
libnetfs/set-get-trans.c:netfs_set_translator() returning EOPNOTSUPP.

Implementing netfs_S_file_set_translator() calling file_set_translator()
in trans/fakeroot.c overrides this call and solves the problem.

Some questions:
- Are the checks in the beginning really needed doesn't
file_set_translator() take care of that?
- Is the explanation really OK?
- AOB?

trans_fakeroot_socket.patch
Description: Text Data