#64 unit test failures in the current master branch

[Andy Tai]

Unit tests fail with the current master branch... have not looked at the details yet just open a ticket for tracking

PASS: import_test
../sqrat.git/autotools/test-driver: line 95: 6872 Segmentation fault "$@" > $log_file 2>&1
FAIL: class_instances
PASS: class_properties
PASS: const_bindings
FAIL: class_binding
PASS: script_loading
../sqrat.git/autotools/test-driver: line 95: 6881 Aborted "$@" > $log_file 2>&1
FAIL: squirrel_functions
PASS: function_overload
../sqrat.git/autotools/test-driver: line 95: 6919 Aborted "$@" > $log_file 2>&1
FAIL: table_binding
../sqrat.git/autotools/test-driver: line 95: 6929 Aborted "$@" > $log_file 2>&1
PASS: null_pointer_return
FAIL: array_binding
PASS: function_params
PASS: sqrat_vm
PASS: suspend_vm
PASS: func_input_argument_type
PASS: run_stack_handling

============================================================================
Testsuite summary for sqrat 0.9
============================================================================

TOTAL: 16

PASS: 11

SKIP: 0

XFAIL: 0

FAIL: 5

XPASS: 0

ERROR: 0

============================================================================

More complete test log attached

[Wizzard]

Some of these failures look like they are due to default constructors no longer being implicitly made.
The errors with "Assertion `m_Ptr != __null' failed" really confuse me though.
I don't have much time currently to take a more in-depth look,
but I will certainly get to it eventually.

Edit: About the default constructors no longer being made, it is now important to call Sqrat::Class::Ctor<>() if one wants a default constructor.

[Andy Tai]

Wizzard, about the oyhter remaining failure (other than the ones due to some shared pointer issue) is a conversion from const char * to string.

I find that change c5b6be6b5786aac91b66d7d8fb113f17b45913c3
is the cause. Is that change really necessary? Seamless conversion from char * to string has always been a feature of Sqrat.

[Wizzard]

If that is the problem, I think I intend things that are already pointers or strings not to
use V as the type for the template and others use const V&
This might be a case for usage of something like std::is_reference and std::is_pointer

[Wizzard]

Fixed in the latest commit

[Andy Tai]

thanks