README.md

In PredictiveEcology/lazyR: Create Lazy Load Databases for R Objects

lazyR

An R package for stashing objects in lazy load databases, analogous to lazy loaded packages. This uses the archivist package for the back end, using the .rda files or raster-based files. The lazy mechanism is at loading time only, creating a promise to evaluate the loadFromLocalRepo() function when it is needed.

Using manual lazy saving

The basic work flow is:

1. decide on a folder for the database, and set it using lazyDir()

2. save objects to a lazyR database using lazySave(obj)

   • optionally give it tags while saving: lazySave(obj, tags=c("maps", "USA"))

3. list the contents of the lazyR database with lazyLs()

4. re-load objects via name or tag using lazyLoad2(tag="maps")
5. remove objects via lazyRm("objName") or lazyRm(lazyLs("tagname"))

Using caching (automatic lazy saving)

An alternative workflow is to use the %<% (assignCache) operator, which will automatically cache the object with its assigned object name. Caching, like with the archivist package, will first do a hash of the arguments in the function, compare with the hashed value of the objects in the database, and return the cached object, if the hash values are the same.

Example:

1. decide on a folder for the database, and set it using

   lazyDir(tempdir(), create=TRUE)

2. assign objects via %<%. See help(cacheAssign)

   a %<% seq(1,10,1)

Common things to use:

• lazyLs() will list all objects in the database
• lazyLs(tagType="all") will list the full archivist data.frame with columns: md5Hash, tag, dateCreated.
• others to come

Connection with archivist package

The lazyDir argument in the lazyR package is exactly the same as a repository in archivist. Thus, all archivist functions work with the lazyR package. Simply use lazyDir and RepoDir interchangeably.

Notes:

• Some tags are done automatically

  • a full set of classes, a result of a call to is(obj), prefixed by class: in the tags
  • the original object name, prefixed with objectName:

• Objects of class Raster* are special because of their "sometimes on disk" nature.

  • If there is a file backing the object, then this file will be copied to the lazyDir directory, with a subdirectory called "raster"
  • The original file will be used as part of the lazy loading.
  • If the original file is a temporary file, be sure to use lazySave(..., copyRasterFile=TRUE) to save it. This is also required to ensure portability of a lazyR database.

PredictiveEcology/lazyR documentation built on May 8, 2019, 3:10 p.m.