The last few days I've been hacking together some more tests and indeed have now read our entire suite of tests. While I was doing this I noticed that we seem to have several ways of doing things so it might be worth deciding on the best.
Firstly the clean up of node.db between tests. We currently have 3 if not 4 ways of doing this.
foreach my $del_node ( $wiki->list_all_nodes ) { print "# Deleting node $del_node\n"; $wiki->delete_node( $del_node ) or die "Can't delete $del_node"; }
which doesnt clear indexes.
Wiki::Toolkit::Setup::SQLite::cleardb( { dbname => "t/node.db" } );
which also doesnt clean indexes
unlink "t/node.db"; unlink <t/indexes/*>; Wiki::Toolkit::Setup::SQLite::setup( { dbname => "t/node.db" } );
which is the most useful. then in one test that becomes a sub sub refresh_db { unlink "t/node.db"; unlink <t/indexes/*>; Wiki::Toolkit::Setup::SQLite::setup( { dbname => "t/node.db" } ); }
and then used several times. Repeatitive code pains me. Therefore I feel we should make refresh_db a sub in OpenGuides::Test and then use that. I have a patch for that and will also probably go through and clean up all the tests.
Secondly, We have 3 ways to do node generation. All of which are fine but we should probably pick one and run with it. This would take more time to change in the test suite. Also there maybe reasons in a test why one way is used above another.
The first way is
OpenGuides::Test->write_data( node => "Wombats", guide => $guide, username => "bob", edit_type => "Minor tidying", return_output => 1, );
which IHMO should be the prefered method.
the second method seem to be along ther lines of
my $q = CGI->new; $q->param( -name => "content", -value => "foo" ); $q->param( -name => "categories", -value => "" ); $q->param( -name => "locales", -value => "" ); $q->param( -name => "phone", -value => "" ); $q->param( -name => "fax", -value => "" ); $q->param( -name => "website", -value => "" ); $q->param( -name => "hours_text", -value => "" ); $q->param( -name => "address", -value => "" ); $q->param( -name => "postcode", -value => "" ); $q->param( -name => "map_link", -value => "" ); $q->param( -name => "os_x", -value => "" ); $q->param( -name => "os_y", -value => "" ); $q->param( -name => "username", -value => "bob" ); $q->param( -name => "comment", -value => "foo" ); $q->param( -name => "edit_type", -value => "Minor tidying" ); $ENV{REMOTE_ADDR} = "127.0.0.1";
my $output = $guide->commit_node( return_output => 1, id => "Wombats", cgi_obj => $q, );
the third method is
$wiki->write_node( "British Museum", "Huge museum, lots of artifacts.", undef, { category => ["Museums", "Major Attractions"] , locale => ["Holborn", "Bloomsbury"] } );
Another thing we need to look at is config generation. we should probably just be using my $config = OpenGuides::Test->make_basic_config;
then changing anything we want with lines like. $config->script_name( "wiki.cgi" ); $config->script_url( "http://example.com/" );
instead of my $config = OpenGuides::Config->new( vars => { dbtype => "sqlite", dbname => "t/node.db", indexing_directory => "t/indexes", script_name => "wiki.cgi", script_url => "http://example.com/", site_name => "Test Site", template_path => "./templates", home_name => "Home", use_plucene => 1 } );
its possibly worth extending the basic config in Openguides::Test.
Finally a couple of the test files have useful subroutines in them whch should probably be put into Openguides::Test.
So thoughts, questions or indeed comments?