prov.run: Provenance Collection Functions

In End-to-end-provenance/rdt: Provenance Collector

Provenance Collection Functions

Description

prov.init intializes a new provenance graph. This function can be executed in the console or placed inside an R script.

prov.save saves the current provenance graph to a prov-json file. If more R statements are executed, the provenance for these statements is added to the graph. The graph is finalized with prov.quit. This function can be executed in the console or placed inside an R script.

prov.quit saves and closes the current provenance graph. This function can be executed in the console or placed inside an R script.

prov.run initiates execution of a script and collects provenance as the script executes. This function should be used if you want to collect provenance for a script that is in an R file and you do not want to modify the R script directly to include calls to prov.init, prov.save and prov.quit. It essentially wraps the execution of the script with calls to prov.init and prov.quit.

prov.source loads an R script and executes it, collecting provenance as it does so. It assumes that provenance has already been initialized, either via a call to prov.init, or because the R script was executed using prov.run. If you want to collect provenance inside scripts that are loaded with R's source function, you should replace calls to source with calls to prov.source.

Usage

prov.init(
  prov.dir = NULL,
  overwrite = TRUE,
  annotate.inside.functions = FALSE,
  first.loop = 1,
  max.loops = 0,
  snapshot.size = 0,
  hash.algorithm = "md5",
  save.debug = FALSE
)

prov.save(save.debug = FALSE)

prov.quit(save.debug = FALSE)

prov.run(
  r.script.path,
  prov.dir = NULL,
  overwrite = TRUE,
  details = TRUE,
  annotate.inside.functions = FALSE,
  first.loop = 1,
  max.loops = 0,
  snapshot.size = 0,
  hash.algorithm = "md5",
  save.debug = FALSE,
  display = FALSE,
  exprs,
  ...
)

prov.source(file, exprs, ...)

Arguments

prov.dir	the directory where the provenance graph will be saved. If not provided, the directory specified by the prov.dir option is used. Otherwise the R session temporary directory is used.
overwrite	if FALSE, includes a time stamp in the provenance graph directory name.
annotate.inside.functions	if TRUE, provenance is collected inside functions.
first.loop	the first loop to collect provenance in a for, while, or repeat statement.
max.loops	the maximum number of loops to collect provenance in a for, while, or repeat statement. If max.loops = -1, there is no limit. If max.loops = 0, no loops are annotated. If non-zero, it indicates the number of iterations of each loop for which provenance should be collected. If max.loops is non-zero, provenance is also collected inside if-statements.
snapshot.size	the maximum size for snapshot files. If 0, no snapshots are saved. If Inf, the complete state of an object is stored in the snapshot file. For other values, the head of the object, truncated to a size near the specified limit, is saved. The size is in kilobytes.
hash.algorithm	the hash algorithm to use for files. Choices are md5 (default), sha1, crc32, sha256, sha512, xxhash32, xxhash64 and murmur32. This feature uses the digest function from the digest package.
save.debug	If TRUE, debug files are saved to the debug directory. This is intended for developers of the rdt / rdtLite package.
r.script.path	the full path to the R script file that is being executed. A copy of the script will be saved with the provenance graph.
details	if FALSE, provenance is not collected for top-level statements.
display	if TRUE, the provenance graph is displayed in DDG Explorer
exprs	Instead of specifying file, an expression, call, or list of call's, can be passed in to be executed.
...	parameters passed on to the source function. See documentation of source for details.
file	the name of the R script file to source.

Details

rdt is an R package that collects provenance as an R script executes. The resulting provenance provides a detailed record of the execution of the script and includes information on the steps that were performed and the intermediate data values that were created. The resulting provenance can be used for a wide variety of applications that include debugging scripts, cleaning code, and reproducing results.

There are two ways in which a user can collect provenance. To collect provenance from commands stored in a script file, use prov.run. This will execute the commands that are in the script, collecting provenance as it does so.

The user can also collect provenance while executing commands in the console. To do this, first execute prov.init. Then enter console commands as normal. When done with the commands for which you want provenance, use prov.quit. If you want to save the current provenance without turning off provenance collection, call prov.save instead of prov.quit. You can call prov.save multiple times before calling prov.quit. Each call will append to the same provenance file.

The provenance is stored in PROV-JSON format. For immediate use it may be retrieved from memory using the prov.json function. For later use the provenance is also written to the file prov.json. This file and associated files are written by default to the R session temporary directory. The user can change this location by (1) using the optional parameter prov.dir in the prov.run or prov.init functions, or (2) setting the prov.dir option (e.g. by using the R options command or editing the Rprofile.site or .Rprofile file). If prov.dir is set to ".", the current working directory is used.

The level of detail collected by rdt may be set using parameters of the prov.run and prov.init functions. Options include collecting provenance inside functions and inside control constructs and saving snapshots of large intermediate values as separate files. These features are turned off by default to optimize performance. Common settings for the level of detail can also be set and managed using the prov.set.detail and related functions.

If prov.source is called when provenance is not initialized, it will just source the file. No provenance will be collected.

Value

prov.init initializes the provenance collector. The prov.init function does not return a value.

prov.save writes the current provenance to a file but does not return a value.

prov.quit writes the current provenance to a file but does not return a value.

prov.run runs a script, collecting provenance as it does so. It does not return a value.

The prov.source function does not return a value.

See Also

prov.json for access to the JSON text of the provenance, prov.visualize to view the provenance graphically. prov.set.detail to see an alternative way to set the amount of provenance collected. prov.annotate.on and prov.annotate.off to see how to control annotation of individual functions

Examples

## Not run: prov.run ("script.R")
prov.init()
a <- 1
b <- 2
prov.save()
ab <- a + b
prov.quit()

End-to-end-provenance/rdt documentation built on Aug. 11, 2022, 12:55 p.m.