Regular expression defining the separation between two input records.
The record as a whole is defined to be‘_.*<record_separator>_’.
When
@nil, the
input is not split into predefined units. The latter may be useful when
the output of a process is dumped in an editor.
Defaults: The default is regex('\n'), which
implies the input if split into (complete) lines.
- See also
- - stream
->record_separator
- process-input_message
Regex that describes the record separator. When not @nil,
the system will collect data from the stream until a complete record is
available before calling the stream<-input_message.
If the data arriving from the stream contains multiple records, the
stream<-input_message
will be called with each record separately.
The default record separator is’\n’(newline),
which implies that input is split into physical lines.
If stream<-record_separator
is an integer, the input is split into records of the indicated size.
This is useful for receiving binary data that is organised in records.
If you just want to collect all data from the process, it is advised
to set the stream<-record_separator
to @nil,
disabling the input filtering.
The stream->record_separator
may be changed on an open connection. Changing the record separator will
immediately scan the data in the input buffer and dispatch any complete
record according to the new definition. Setting the separator to @nil
will run the stream<-input_message
using the data in the buffer and destroy the buffer.
