v0.18.0
Breaking changes:
- functions that create a new
DataFrame
copy passed columns by default; this can be overridden bycopycols
keyword argument or by using theDataFrame!
function that does not copy passed columns - in
eachcol
nownames
keyword argument defaults tofalse
- make
dropmissing
anddropmissing!
disallow missing values in column by default - removed long deprecated uses of
by
,nullable!
,keys
,values
,pool
,pool!
,complete_cases
,complete_cases!
,sub
,rename!
,rename
andvcat
- removed dependency on DataStreams.jl and WeakRefStrings.jl
Deprecations:
- several
DataFrame
constructors not taking source data were deprecated colwise
is now deprecated
Enhancements:
- add
cols
keyword argument tovcat
, make it ignore data frames with no columns, and support efficientreduce
- allow passing a data frame with no columns to
append!
- allow
push!
to a data frame with no columns and addcols
keyword argument to it - improvements of showing data frames, grouped data frames and data frame rows for CSV, TSV, HTML and LaTeX MIME types
- optimized grouping methods for
PooledArrays
DataFrame
constructor can now take tuples of column vectors and column names- added
compress
keyword argument to thecategorical!
function describe
now supports passing custom functions- make
allowmissing!
anddisallowmissing!
accept vector ofBool
- add
select
,select!
anddeletecols
functions
Bug fixes:
combine
now has a better handling of combining incompatible return values
Miscellaneous:
- started testing against Julia 1.1 and stopped against Julia 0.7
- migration to JuliaRegistrator
- numerous documentation improvements
- removed custom
deepcopy
implementation - removed dependency on CodecZlib,jl, TranscodingStreams.jl
- improved error message when column is not found in a data frame