Some test coverage (combinatorics and intfuncs) #9620
Conversation
| @test binomial(5,3) == 10 | ||
| @test binomial(2,1) == 2 | ||
| @test binomial(1,2) == 0 | ||
| @test binomial(-2,1) == -2 |
There was a problem hiding this comment.
I guess this makes sense, but I found it surprising. Add a comment to make sure someone doesn't delete it out of puzzlement later?
There was a problem hiding this comment.
I was adding some edge cases this seems to agree with Wolfram Alpha!
|
Awesome! |
|
Have an int32 failure: I did wonder if that might be an issue when I was writing it. Will post a fix, hopefully that's the only one. |
|
|
||
| @test base(2, 5, 7) == "0000101" | ||
|
|
||
| @test bits(1035) == "0000000000000000000000000000000000000000000000000000010000001011" |
There was a problem hiding this comment.
will this be different on int32 also? Edit: yes it will.
|
@timholy Thanks for you comments, updated (and hopefully will pass appveyor now). Edit: it did. |
|
(rebased over d877d63) |
|
3/4 passes and one totally random-seeming failure. We must get to the bottom of these random failures, but I think adding more tests is a good step in that direction. So, merging. |
Some test coverage (combinatorics and intfuncs)
|
Many thanks! |
|
may as well also backport these, with the exception of the base64 one which I don't think applies to release-0.3 |
(cherry picked from commit ff0470c) ref PR #9620 Conflicts: test/combinatorics.jl TST intfuncs more coverage (cherry picked from commit 3d376d1) (leave out ndigits tests that rely on #8266 which was not backported) Conflicts: test/intfuncs.jl TST fix 32bit bits test, add comment to interesting binomial result (cherry picked from commit 93f6ccb)
|
Backporting the new tests revealed an interesting difference in behavior between 0.3 and 0.4 for |
(Just a couple of low-hanging test files, regression tests.)
cc #9493