Omit parent key in tp/t test suite?

[Gavin Smith]

Hello Patrice, what would you think about omitting the 'parent' pointers
in the test suite under tp/t?  Here is the patch:

diff --git a/tp/t/test_utils.pl b/tp/t/test_utils.pl
index d5ca1e1359..6a585f1d4e 100644
--- a/tp/t/test_utils.pl
+++ b/tp/t/test_utils.pl
@@ -471,15 +471,14 @@ my @sections_keys = ('section_next', 'section_prev', 
'section_up',
 my @node_keys = ('node_next', 'node_prev', 'node_up', 'menus',
   'associated_section', 'node_preceding_part');
 my %avoided_keys_tree;
-my @avoided_keys_tree = (@sections_keys, @menus_keys, @node_keys, 'structure',
-   'menu_child', 'unit_next', 'directions', 'page_next', 'remaining_args');
+my @avoided_keys_tree = (@sections_keys, @menus_keys, @node_keys,
+    'structure', 'menu_child', 'unit_next', 'directions', 'page_next',
+    'remaining_args', 'parent');
 foreach my $avoided_key(@avoided_keys_tree) {
   $avoided_keys_tree{$avoided_key} = 1;
 }
 sub filter_tree_keys { [grep {!$avoided_keys_tree{$_}} ( sort keys %{$_[0]} )] 
}
 
-#my @avoided_compare_tree = (@avoided_keys_tree, 'parent', 'node_tree');
-
 my %avoided_keys_sectioning;
 my @avoided_keys_sectioning = ('section_next', @contents_keys, @menus_keys,
   @node_keys, 'menu_child', 'manual_content', 'toplevel_next');


These pointers generate a huge amount of non-human readable "fix up"
code from Data::Dumper, after the legible tree dump.  Updating these test
results gets rid of these.  For example:


@@ -154,29 +139,12 @@ $result_trees{'alias_table_command'} = {
           }
         }
       ],
-      'parent' => {},
       'type' => 'before_node_section'
     }
   ],
   'type' => 'document_root'
 };
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[0]{'args'}[0]{'contents'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[0]{'args'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[0]{'args'}[0]{'parent'}
 = $result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[0]{'parent'} = 
$result_trees{'alias_table_command'}{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[1]{'parent'} = 
$result_trees{'alias_table_command'}{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'args'}[0]{'contents'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'args'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'args'}[0]{'parent'}
 = $result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0]{'contents'}[0]{'args'}[0]{'contents'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0]{'contents'}[0]{'args'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0]{'contents'}[0]{'args'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0]{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0]{'contents'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'contents'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[0]{'parent'}
 = $result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[1]{'args'}[0]{'contents'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[1]{'args'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[1]{'args'}[0]{'parent'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[1];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'contents'}[1]{'parent'}
 = $result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2];
 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'extra'}{'command_as_argument'}
 = 
$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'args'}[0]{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'contents'}[2]{'parent'} = 
$result_trees{'alias_table_command'}{'contents'}[0];
-$result_trees{'alias_table_command'}{'contents'}[0]{'parent'} = 
$result_trees{'alias_table_command'};
 
 $result_texis{'alias_table_command'} = '@alias myalias = code
 

It also reduces the size of the distribution .tar.xz file (roughly from
5.2 MB to 4.9 MB), and saves 10s of MB when extracting and building.

It should also make the test suite run faster, due to less code being loaded
and parsed by Perl:

Before:

  $ time make check
  real    2m43.365s
  user    2m27.531s
  sys     0m13.236s
  ...
  $ time make check
  real    2m45.350s
  user    2m27.508s
  sys     0m13.864s

After:

  $ time make check
  ...
  real    2m37.472s
  user    2m19.646s
  sys     0m13.453s
  $ time make check
  ...
  real    2m41.970s
  user    2m20.703s
  sys     0m13.966s

(These times also include tp/tests, which is not affected by
this change.)

Testing an individual test file:

Before:

  $ time perl -w t/converters_tests.t
  ...
  real    0m5.692s
  user    0m5.536s
  sys     0m0.088s
  $ time perl -w t/converters_tests.t
  ...
  real    0m5.598s
  user    0m5.474s
  sys     0m0.104s

After:

  $ time perl -w t/converters_tests.t
  ...
  real    0m5.153s
  user    0m4.947s
  sys     0m0.105s
  $ time perl -w t/converters_tests.t
  ...
  real    0m5.029s
  user    0m4.918s
  sys     0m0.092s

This is about 5-10% quicker and may be more significant on systems
with slower storage devices and/or CPU's.

Do you have any thoughts or objections before I commit this?