cross-partition mv fails with Operation not supported due to ACLs
FYI, on a system like this, $ head -1 /etc/issue Red Hat Enterprise Linux AS release 4 (Nahant Update 6) $ rpm -q glibc-devel glibc-devel-2.3.4-2.39 $ uname -r 2.6.9-67.ELsmp The coreutils test, tests/mv/part-symlink, fails with this output: -0 mv rem_reg loc_sl (loc_sl) () -0 mv -b rem_reg loc_sl (loc_sl loc_sl~ - rem_reg) () +0 mv rem_reg loc_sl [mv: preserving permissions for `loc_sl': Operation not supported ](loc_sl) () +0 mv -b rem_reg loc_sl [mv: preserving permissions for `loc_sl': Operation not supported ](loc_sl loc_sl~ - rem_reg) () In spite of the name of that test, the failure is unrelated to symlinks. Any cross-partition move of a regular file elicits the Operation not supported diagnostic, *but*, and here's the part that matters: It happens only when moving from an NFS-mounted partition (which is Solaris 10 ZFS, I think) to a partition without those newer ACLs. E.g., when moving to an ext3 /tmp or to /dev/shm, which is tmpfs, there's no problem, because acl_*get*_fd fails. Investigating shows that the diagnostic is due to acl_set_fd failing, which is because it does this: fsetxattr(4, system.posix_acl_access..., \x02..., 36, 0) = -1 EOPNOTSUPP (Operation not supported) Since coreutils-5.2.1 works the same way on that system, this isn't even a regression. ___ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils
Re: inconsistency in seq
Vincent Rossetto wrote: Hi I have the coreutils version 5.97. I wonder why the behaviour of seq is different for integer and non-integer INCREMENT. Examples: $ seq 10 10 30 10 20 30 $ seq 0.1 0.1 0.3 0.1 0.2 $ According to the man page, it should be the same result namely the first example, which is correct. The info page for your version should give you more info about this floating point related issue. Actually this issue was fixed in the recently released version 6.9.90 http://git.sv.gnu.org/gitweb/?p=coreutils.git;a=blob;f=NEWS coreutils releases tend to be slow about propagating to distributions for various reasons unfortunately. Pádraig. ___ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils
inconsistency in seq
Hi I have the coreutils version 5.97. I wonder why the behaviour of seq is different for integer and non-integer INCREMENT. Examples: $ seq 10 10 30 10 20 30 $ seq 0.1 0.1 0.3 0.1 0.2 $ According to the man page, it should be the same result namely the first example, which is correct. Greetings, Vincent -- Vincent Rossetto Laboratoire de physique et modélisation de la matière condensée / CNRS Grenoble mailto:[EMAIL PROTECTED] phone: (+33) 4 56 38 71 35. ___ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils
