Re: OpenEuphoria and the tests?
- Posted by SDPringle Oct 11, 2017
- 1137 views
I fixed the test fail in t_declasgn_wrning.e by adding two more test_equals because they are in the warning.lst file generated by t_declasgn.e
I do not see such a change. Do you think you could put your changes in a paste bin, so it can be incorporated? I am not sure how you fixed these tests.
With so many platforms, architectures and compilers, does openeuphoria produce the same warnings on all, all the time?
They are supposed to.
Do the tests keep up with code changes like 4.0.5 to 4.1.0?
No. The tests and the code and library code go into the same development branch. So, if a test changes between 4.0.5 and 4.0.6, the test will be as it was when you revert your code snapshot to 4.0.4. The design is tests written for 4.0 gets merged into 4.1. Tests that do not apply to 4.0 are put into 4.1 directly.
That only leaves me t_callc.e, t_literals.e and t_socket.e failing some tests.
I think t_literals.e is related to t_callc.e by MATH or assembly registers:
Is this the same problem I remember looking at where the difference between values computed on Sun and Intel chips differ by the least significant bit in some cases? I guess t_callc.e could give you an error because of these differences in math. Are we talking about the same kind of difference?
SD Pringle
