Re: [PATCH] sort: Improve sort --random-sort test.

[Jim Meyering]

address@hidden (Bob Proulx) wrote:
> * tests/misc/sort-rand: If "locale" is available pick a random
> non-C locale and check "sort --random-sort" behavior using it.
>
> Signed-off-by: Bob Proulx <address@hidden>
> ---
>  ChangeLog            |    6 ++++++
>  tests/misc/sort-rand |   13 +++++++++++++
>  2 files changed, 19 insertions(+), 0 deletions(-)
>
> diff --git a/ChangeLog b/ChangeLog
> index d5be794..327e534 100644
> --- a/ChangeLog
> +++ b/ChangeLog
> @@ -1,3 +1,9 @@
> +2007-07-26  Bob Proulx  <address@hidden>
> +
> +     sort: Improve sort --random-sort test.
> +     * tests/misc/sort-rand: If "locale" is available pick a random
> +     non-C locale and check "sort --random-sort" behavior using it.
> +
>  2007-07-24  Jim Meyering  <address@hidden>
>
>       sort: add a test to exercise the affected code.
> diff --git a/tests/misc/sort-rand b/tests/misc/sort-rand
> index 0380af8..78f48ef 100755
> --- a/tests/misc/sort-rand
> +++ b/tests/misc/sort-rand
> @@ -49,4 +49,17 @@ # Fail if the sorted output is not the s
>  sort -n out > out1
>  cmp in out1 || { fail=1; echo "not a permutation" 1>&2; }
>
> +# If locale is available then use it to find a pseudo-random non-C locale.
> +if (locale --version) > /dev/null 2>&1; then
> +  locale=`locale -a | sort --random-sort | awk '/^.._/{print;exit}'`
> +  LC_ALL=$locale sort --random-sort in > out || fail=1
> +
> +  # Fail if the input is the same as the output.
> +  cmp in out > /dev/null && { fail=1; echo "not random with LC_ALL=$locale" 
> 1>&2; }
> +
> +  # Fail if the sorted output is not the same as the input.
> +  sort -n out > out1
> +  cmp in out1 || { fail=1; echo "not a permutation with LC_ALL=$locale" 
> 1>&2; }

Thanks, Bob.
However, note that even the buggy sort -R *does* change the order.
It's just that it always does it the same way :-)
(this is on a rawhide system)

  $ (export PATH=/bin:$PATH LC_ALL=fr_FR.utf8; seq 25|sort -R)
  1
  10
  11
  12
  13
  14
  15
  16
  17
  18
  19
  2
  20
  21
  22
  23
  24
  25
  3
  4
  5
  6
  7
  8
  9

so if your test were to run sort -R twice, and compare *those* outputs,
then it would trigger on this bug.