bdt.eval: big.data.table evaluate on nodes

In jangorecki/big.data.table: Parallel distributed computing for data.table

Description

Main engine for passing queries to nodes, control parallelization, rbinding or simplifing returned object. Allows to measure timing and verbose messages.

Usage

bdt.eval(x, expr, lazy = TRUE, send = FALSE, simplify = TRUE,
  rbind = TRUE, parallel = TRUE, silent = TRUE, outer.aggregate = FALSE,
  .log = getOption("bigdatatable.log", FALSE))

bdt.eval.log(x, expr, lazy = TRUE, send = FALSE, simplify = TRUE,
  rbind = TRUE, parallel = TRUE, silent = TRUE, outer.aggregate = FALSE,
  .log = getOption("bigdatatable.log", FALSE))

Arguments

x	big.data.table.
expr	expression.
lazy	logical if TRUE then *expr* is substituted.
send	logical, if TRUE submit expression appended with 'TRUE' to not fetch potentially big results from provided *expr*, useful for data.table *set** or ':=' functions.
simplify	logical if *TRUE* (default) it will simplify list of 1 length same type objects to vector.
rbind	logical if *TRUE* (default) results are data.table they will be rbinded.
parallel	logical if parallel *TRUE* (default) it will send expression to nodes using 'wait=FALSE' and collect results afterward executing each node in parallel.
silent	should be silently catched by 'try' or 'logR' if enabled.
outer.aggregate	logical or a function, if *TRUE* will able the same query to rbind of results from each node, should not be used with '.SD', '.N', etc. Also conflicts with filtering in 'i'. Can be also a function taking first argument 'x' rbinded data.table.
.log	logical if *TRUE* then logging will be done using logR to postgres db.

Details

The 'bdt.eval.log' function is wrapper on 'bdt.eval' with 'logR()' call, so use only when logR connected.

Value

Depending on *simplify, rbind* the results of evaluated expression.

jangorecki/big.data.table documentation built on May 18, 2019, 12:24 p.m.