The APPENDICES
directory of the Raku specification contains advisory
specifications. These tests describe behaviour some implementations chose
to follow and other implementations may follow the same behaviour, to offer
similar execution results from the same Raku programs.
However, it's possible some implementations or some environments may not
be able to adhere to these specifications. For example, an implementation
that can compute 2**-10000000000
sufficiently fast may choose to use
higher limits for powers before throwing overflow exceptions.
In summation, the APPENDICES contain tests that fall somewhere between implementation-specific tests residing in test suites of particular implementations and regular specification tests that all implementations must adhere to.
An implementation may fail all of the tests in APPENDICES, yet still claim the status of being a fully-compliant Raku implementation.
Contains tests of potentially-desirable behaviour that's dependent on the limitations of typical environments. For example, raising a number to a huge power effectively "hangs" the program, so the tests check such cases throw an overflow error instead.
This appendix contains things that may one day provide useful behaviours, but
at the moment they don't, yet they have to exist in implementations for one
reason or another (throwing an X::NYI
exception). For example, one day
we might come up with a very useful usecase for write methods of an
IO::CatHandle
, but until that day, these methods must throw an X::NYI
,
because otherwise the versions of the parent's (IO::Handle
) would be used
instead, providing confusing behaviour.
Sometimes during the creation of the next version of the spec various bugs get fixed in features deprecated in fewer specs but are still available in older language versions. This appendix contains such tests. It's possible newer implementations may choose not to implement features that are already deemed deprecated in the oldest version of the specification those implementations decide to implement.
Some behaviour is currently experimental and hasn't been fully agreed on. The tests in this appendix exercise that behaviour either directly or indirectly. There's a high chance these tests will change in the future.