Closed bkamins closed 2 years ago
Merging #47 (35723d7) into main (3e905a6) will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## main #47 +/- ##
=======================================
Coverage 95.12% 95.12%
=======================================
Files 1 1
Lines 41 41
=======================================
Hits 39 39
Misses 2 2
Impacted Files | Coverage Δ | |
---|---|---|
src/DataAPI.jl | 95.12% <ø> (ø) |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 3e905a6...35723d7. Read the comment docs.
@nalimilan - now I realized that the approach proposed here (and discussed on slack) cannot be used.
The reason is that we cannot dispatch on DataFrame
since its type is DataType
(it is not a function).
Maybe we should leave https://github.com/JuliaData/DataFrames.jl/pull/3031 as it is then? Or do you see any other solution? The only I can see is that allcombinations
would define a table-like type that then would be passed to DataFrame
constructor. But then we would need to provide implementation of the logic of the function in DataAPI.jl, which I think we should not do.
Can't we just define methods like allcombinations(::Type{DataFrame}, ...)
?
Ah - right. We can 😄. In this case can you please review this PR?
@quinnj We could mention in the docstring that TableOperations owns allcombinations(NamedTuple, ...)
and implement a method there.
@quinnj - could you have a look at this PR before I merge it and make a release (and finish https://github.com/JuliaData/DataFrames.jl/pull/3031 a follow-up PR in DataFrames.jl)? Thank you!
Thank you!
add
allcombinations
function. see https://github.com/JuliaData/DataFrames.jl/pull/3031 for an example implementation.