Part of the parallelization improvement plan tracked in PARALLELIZATION.md (Priority 4, optional/advanced).
Split out of #16 so it can be scoped and closed independently of backend selection and core-count heuristics.
Background
af(), vis(), and bglmnet() all run furrr::future_map() bootstrap loops that can take anywhere from a few seconds to several minutes depending on B/n.c/nlambda, with no feedback to the user while running.
Tasks
See PARALLELIZATION.md section "Performance Considerations" and the Phase 4 checklist for context.
Part of the parallelization improvement plan tracked in PARALLELIZATION.md (Priority 4, optional/advanced).
Split out of #16 so it can be scoped and closed independently of backend selection and core-count heuristics.
Background
af(),vis(), andbglmnet()all runfurrr::future_map()bootstrap loops that can take anywhere from a few seconds to several minutes depending onB/n.c/nlambda, with no feedback to the user while running.Tasks
progressrsupport toaf()'sfuture_map()call (then.c-value bootstrap loop)progressrsupport tovis()'sfuture_map()call (theB-replicate loop)progressrsupport tobglmnet()'sfuture_map()call (theB-replicate loop)progressrtoSuggests(orImports, depending on default-on/off decision) inDESCRIPTIONprogressrhandler registered (and with none registered)See PARALLELIZATION.md section "Performance Considerations" and the Phase 4 checklist for context.