Re: [PATCH] tests: fix a failure when install-info is not available

[Stefano Lattarini]

Hi Dmitry, thanks for the patch.  But I'm not sure it's right (see below).

On 04/13/2012 02:58 AM, Dmitry V. Levin wrote:
> This fixes a regression introduced by commit v1.11-766-g619c1b2.
> 
> * tests/install-info-dir.test: Do not attempt to read or modify
> $instdir/info/dir file when install-info is not available.
>
> Signed-off-by: Dmitry V. Levin <address@hidden>
> ---
>  tests/install-info-dir.test |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/tests/install-info-dir.test b/tests/install-info-dir.test
> index 405480d..1b75e8d 100755
> --- a/tests/install-info-dir.test
> +++ b/tests/install-info-dir.test
> @@ -155,16 +155,16 @@ for val in no NO n; do
>    test -f $instdir/info/foo.info
>    test ! -f $instdir/info/dir
>  done
>  
> +if test $have_installinfo = yes; then
>
>  $MAKE install-info
>  chmod a-w $instdir/info/dir
>  for val in no NO n; do
>    env AM_UPDATE_INFO_DIR="$val" $MAKE uninstall
>    $FGREP 'Does nothing at all, but has a nice name' $instdir/info/dir
>  done
>
Hmm... actually, the fact that the above was run also when 'install-info'
might not be available was done *on purpose*, because the exporting of
'AM_UPDATE_INFO_DIR' to a "no" value should prevent 'install-info' from
being run at all.

Have this patch motivated by a real failure (in which case, I'd like to
read details about it), or just by the fact that the test looked wrong?
In the second case, I'd drop the patch, and instead add an explicative
comment to the test repating what I've explained above.

Thanks,
  Stefano