Note that if you don't want links in your paths (file.dir
, work.dir
) to get resolved and have
complete control over the way the path is used internally, pass an absolute path which begins with “/”.
makeRegistry(
id,
file.dir,
sharding = TRUE,
work.dir,
multiple.result.files = FALSE,
seed,
packages = character(0L),
src.dirs = character(0L),
src.files = character(0L),
skip = TRUE
)
[character(1)
]
Name of registry. Displayed e.g. in mails or in cluster queue.
[character(1)
]
Path where files regarding the registry / jobs should be saved.
Default is “<id>-files” in current working directory if id
is set.
[logical(1)
]
Enable sharding to distribute result files into different subdirectories?
Important if you have many experiments.
Default is TRUE
.
[character(1)
]
Working directory for R process when experiment is executed.
Default is the current working directory when registry is created.
[logical(1)
]
Should a result file be generated for every list element of the
returned list of the job function?
Note that the function provided to batchMap
or
batchReduce
must return a named list if this is set to TRUE
.
The result file will be named “<id>-result-<element name>.RData”
instead of “<id>-result.RData”.
Default is FALSE
.
[integer(1)
]
Start seed for experiments. The first experiment in the registry will use this
seed, for the subsequent ones the seed is incremented by 1.
Default is a random number from 1 to .Machine$integer.max/2
.
[character
]
Packages that will always be loaded on each node.
Default is character(0)
.
[character
]
Directories containing R scripts
to be sourced on registry load (both on slave and master).
Files not matching the pattern “\.[Rr]$” are ignored.
Useful if you have many helper functions that are needed during the execution of your jobs.
These files should only contain function definitions and no executable code.
Default is character(0)
.
[character
]
R scripts files
to be sourced on registry load (both on slave and master).
Useful if you have many helper functions that are needed during the execution of your jobs.
These files should only contain function and constant definitions and no long running, executable code.
These paths are considered to be relative to your work.dir
.
As a last remedy in problematic cases you can use absolute paths, by passing paths that
start with “/”, see the comment about file.dir
and work.dir
above,
where we allow the same thing.
Note that this is a less portable approach and therefore usually a less good idea.
Default is character(0)
.
[logical(1)
]
Skip creation of a new registry if a registry is found in file.dir
.
Defaults to TRUE
.
[Registry
]
Every object is a list that contains the passed arguments of the constructor.
# NOT RUN {
reg = makeRegistry(id = "BatchJobsExample", file.dir = tempfile(), seed = 123)
print(reg)
# }
Run the code above in your browser using DataLab