- added
compute()
function - misc fixes (mostly
flow_dfr
andflow_dfg
) and more tests - fix
flow_dfr
andflow_dfg
for factors
- the function supplied to
flow_dfr
receives only changed rows and no index - new function:
flow_dfg
hadles grouped data frames - refactor fn_key formation
- misc fixes
- extensive rewrite, breaks compatibility with previous versions
- add: separate cache from eddy (new R6 class tree)
- updated eddy functions
- R6 class tree for flow functions
flow_
functions for file source and namespace sinkflow_dfr
for caching row-wise data frame operations- alternative setup that does not require
make_flow_fn
functions - lazy computation (does not re-compute a pipeline if not needed)
- extensive testing
- add: is_cached to test whether a state already exists in cache
- add: delete_cache to delete the given state from rflow and eddy
- fix: fail to init rflow when no data on disk
- fix: fail when retrieving an rflow element inside cached function
- add:
element()
function for R6Flow objects
[
method for R6Flow to extract its elements
- fix: make_ns_sink changed to follow make_sink pattern
- fix: eddies always create cache_env and dir on disk
- fix: eddies require registered rflow when adding data
- added file source
- add multi-purpose sink
- collect is now an S3 method, using the S3 generic from dplyr
- new functions: new_eddy, get_eddy, delete_eddy
- added tests
- added R6 obj printing
- update documentation
- update memory and disk caching issues
First release