If the YAML tests can represent everything which can be done with the
JSON tests, then as there is a YAML module for python it would be
possible to look at porting the JSON tests to YAML and creating a python
test system to use the YAML data to test that use through the python
bindings is correct. In this way the JSON tests could be removed.
Michael Whapples
On 03/03/2016 09:33, Christian Egli wrote:
Michael Gray <mgray@xxxxxxx> writes:
I should note that I do not use the YAML tests as they do not have theI think I enhanced the YAML tests so that they could handle all the
ability to represent all the emphasis that are used.
emphasis.
I also assume that to run the tests on Windows' Microsoft SDK and YAMLI understand where you are coming from. Let me just shed some light on
as nothing is easy dealing with the Microsoft SDK. The lou_compare
tool was written to deal with these issues and can easily be modified
for whatever testing can be done.
the motivation for the YAML tests. We had 3 different ways to test
things before (JSON tests, Python docests and C based tests), that only
run in certain environments (JSON only worked with UCS4). Our hope with
the YAML tests was that we could consolidate this somewhat. That is why
there is a reluctance towards yet another test framework. But there
should be room for peaceful coexistence :-). What would be nice if it
could be used in `make check`, i.e. it returns true or false, so we can
check for regressions.
Thanks
Christian
