51
Pipeline
A Pipeline object contains a set of bufs and procs. When a pipeline is
executed, it repeatedly executes the procs, which read data from and
to bufs, until no procs can be executed (typically because all of the
data has been processed).
Constructing a
Pipeline
The typical way to construct a pipeline, followed in the example
pipeline-construction functions, is to first create the bufs. Next, the
procs in the pipeline are created. The inputs and outputs of the procs
are specified as bufs, essentially "wiring together" the procs into a
pipeline. Finally, the bufs and procs are encapsulated into a pipeline
object, which can be executed.
Object
Initialization
The basic model for constructing an object is to create it, and then set
various properties to configure it, and then to initialize it. During
initialization, any storage allocation needed before executing the
pipeline is done, according to the set properties.
Proc and Buf objects need to be initialized before they are executed
within a pipeline. Therefore, when they are created, their error string
is set to
"uninitialized". The pipeline execution methods will not
use any procs or bufs with error strings set, so this prevents executing
the pipeline before it is initialized.
Object Name Each object can optionally have a name string. This name is used to
identify the object in status messages.
Error String Each object can have an associated error string. Normally, if no error
has occurred, the error string is the empty string
"". If an error occurs
while initializing or running an object, its error string is set to a string
describing the error. The only way to clear an error is to initialize the
object.
Column Types A buf object, described below, represents a data buffer with rows and
columns, something like an Spotfire S+ data frame. As with a data
frame, different columns can store different types of data. Currently,
there are four different data types: double, factor, string, and
timeDate.
Double
The double type is simple; each value is interpreted as a floating point
number.
Commentaires sur ces manuels