On 03/26/2014 01:26 PM, Bernhard Voelker wrote:
> On 03/25/2014 07:25 AM, Paul Eggert wrote:
>> (the second is just a code cleanup).
> 
> Oops, that 2nd patch
>   http://git.sv.gnu.org/cgit/coreutils.git/commit/?id=7f669b40
> now produces the following check failure:
> 
> FAIL: tests/mv/i-3
> ==================

That's an interesting one. I guess our CI didn't flag this
as the test is skipped without a controlling tty:
http://hydra.nixos.org/jobset/gnu/coreutils-master

Also the test is skipped with a parallel make check!
I.E. this first command detects the fail, while the second skips the test:
  $ make -j2 check TESTS="tests/mv/i-3.sh"
  $ make -j2 check TESTS="tests/mv/i-2.sh tests/mv/i-3.sh"

So to handle both cases I'm updating the "release tasks" list to:
  setsid make check $(($(nproc) * 2)) # tests without controlling tty
  make check -j1 # tests with controlling tty

> BTW: in check_overwrite_prompt, the file 'out' contains
>   "mv: unwritable 'g' (mode 0000, ---------); try anyway?"

That should have been apparent in the test log immediately,
without you having to add debug.

In general we should allow these patches a little
while for review before pushing.

Hopefully the attached fixes up these issues.

thanks,
Pádraig.