Roadmap for v0.3.0 release

[junyuan-chen]

The upcoming v0.2.3 release is very likely to be the last one in the v0.2.x series. Development will target toward the v0.3.0 release once v0.2.3 is out. The goal is to get close to a full coverage of the ReadStat API for complete functionality while maintaining the performance advantage over the related packages.

Here is a list of changes that have been planned for v0.3.0:

• Use MappedArray to lazily convert variables in date/time format back and forth between format-specific numerical values and Julia Date/DateTime. (Support writing date/time values #36)
• Support user-specified missing values (.a, ..., .z and value ranges) while maintaining the ease of getting back to missing (possibly with StructArray and MappedArray).
• Elements in LabeledArrays that do not have a value label should be returned as the original value without being converted to String.
• Support READSTAT_TYPE_STRING_REF for very long strings in Stata.
• Use the integer column fields in ReadStatColumns for the reference arrays of PooledArray.
• Support read/write value labels from/to .sas7bcat files.
• Add an copycols option to writestat for converting the user-provided table to ReadStatTable{ReadStatColumn} before row iteration. This could make writing data files faster. (Support writing date/time values #36)
• Support reading/writing data by chunk.
• Reproduce the sample data used for testing, ideally with scripts executed on Stata/SAS/SPSS.
• Require Julia v1.6 and ReadStat_jll.jl v1.1.8. Check compatibility of dependencies.