Add extra coverage testing for char.jl #12882
Conversation
ScottPJones
force-pushed
the
spj/testchar
branch
2 times, most recently
from
516d48f
to
519b05f
Compare
|
Test failure due to #11553. |
|
Yes, it's killed basically every PR of coverage tests I've submitted this weekend, sometimes more than once. |
ScottPJones
force-pushed
the
spj/testchar
branch
from
519b05f
to
ad657f0
Compare
| # This gets a deprecation warning that seems incorrect | ||
| # getindex('a',1.0,1) -> WARNING: indexing with non Integer Reals is deprecated | ||
| # Capture spurious deprecation warning | ||
| let |
There was a problem hiding this comment.
Again, please remove this. You can leave a note to update when the implementation changes, but for now just test what is currently implemented. This is not going to change for 0.4.
There was a problem hiding this comment.
This is part of 0.4, implemented in line 22 of char.jl:
getindex(c::Char, I::Real...) = getindex(c, to_indexes(I...)...)
|
Please squash these commits. |
Handle deprecation warnings treated as errors Remove getindex check
ScottPJones
force-pushed
the
spj/testchar
branch
from
ad657f0
to
f0194e7
Compare
|
@jakebolewski Squashed, and I removed the test of the getindex/checkbounds that depend on the deprecated |
|
Bump, these have passed the now OOM free tests (yeah!), should be all set. |
Add extra coverage testing for char.jl
Add tests for getindex, bswap, ndims, size and typemin
Note: I noticed a number of inconsistencies that should probably be dealt with in a post-0.4 PR.
getindex('c',1,1,1)is allowed, and returns'c', butgetindex("c",1,1,1)gets an error.bswapon aCharshould probably not be allowed, the operation only makes sense on the underlying codeunit, i.e.UInt32, not onChar.