Dumpsterl interactive probe shell;
start it via typing `ds:shell().' into the Erlang shell.


Commands:

gui [<dumpfile>]
  Launch gui reading the currently configured dump file,
  or the dump file specified by the argument, if present.

help [<keyword>]
  Display help for dumpsterl commands, params and options.
  Type:
  - 'help' to get a list of valid commands;
  - 'help <keyword>' to get help specific to a keyword;
  - 'help commands' to get help for all valid commands;
  - 'help params' to get help for all valid params;
  - 'help options' to get help for all valid options;
  - 'help all' to print all help output.

quit
  Quit the Dumpsterl shell, giving control back to the Erlang shell.
  The probe, if running, will be stopped cleanly.

run
  Run the probe with current parameters and options.

set <param|option> <value>
  Set given param or option to supplied value.
  Type 'help params' to get help for all valid params.
  Type 'help options' to get help for all valid options.

show [<param|option> ...]
  Show currently configured parameter and option values.
  Issued without arguments, 'show' will print a table with all
  configuration values. When given a list of param or option names,
  output will be restricted to those only.

stop
  If the probe is running, stop it cleanly. Even if the probe is stopped
  this way before reaching its set `limit' or end of table, the spec will
  be properly closed, so it will contain all information corresponding to
  the final reported record count.


Params:

attrs
  Attribute selector list to specify metadata attributes.
  Syntax: [{Attr, FieldSelector}, ...] where FieldSelector selects
          the source of data to use for attribute Attr.
          See param `field' for the syntax of FieldSelector.
  Attributes currently utilized by the gui:
  - 'key': Record key to associate with data from the record;
  -  'ts': Timestamp to associate with data from the record.

field
  Record/tuple field to select for inspection. Supported formats:
  - 0 or []: spec the whole record;
  - pos_integer(): spec the selected field;
  - [pos_integer()]: list of chained field references to select data
       from embedded tuples or records. E.g. [3,2] is interpreted as:
       ``take the third field from the record read from the table, then
         take the second field of it to be processed by dumpsterl.''
  If the referenced field does not exist, the record is skipped from the
  spec (but counts toward the limit and shows up in the output count).

table
  The table to read. Its accepted format depends on the `type' setting:
  - mnesia or ets table name (atom);
  - ets table id (integer);
  - dets table name (any term) for an already opened dets table;
  - filename of a dets table or disk_log file.

type
  Type of table. This selects the table driver for the probe.
  Valid types: dets | disk_log | ets | mnesia


Options:

dump: filename() | 'false'
  Dump the accumulated spec on each progress update, but at most once
  every ten seconds (if progress output is enabled), and at the end.
  - 'false': no dump is written;
  - filename(): the accumulated spec is dumped to this file.
  NB. in case of parallel probe execution (see `procs' option), only
  the master process dumps its spec. With procs=N, only 1/N of the
  whole spec is periodically dumped. Naturally, the entire result
  is collected and dumped on completion (or in response to `stop').

hll_b: 4..16 | 'false'
  Parameter 'b' (base 2 log of substream count) of the hyperloglog
  cardinality estimator. Larger values exponentially increase the
  storage requirement and marginally improve estimation precision.
  A setting of 'false' disables cardinality estimation.

limit: pos_integer() | 'infinity'
  Limit the number of records to process. In case reads are done in
  chunks of several records (such as with disk_logs), the limit might
  be slightly exceeded. Defaults to 1000, which is a safely low value.
  A setting of 'infinity' enables processing the whole table.

mnesia_dir: dirname() | 'undefined'
  The name of the mnesia directory where table data files are stored.
  This option is useful if the Erlang node running dumpsterl does not
  run a Mnesia instance, but has access to the database filesystem.

procs: pos_integer()
  Number of parallel processes to use when running the probe.
  Defaults to the number of logical processors as reported by ERTS.

progress: number() | 'false'
  Output progress information and (if dump is enabled) write interim
  dumps.
  - 'false': no output
  - T: update progress info every T seconds
       (achieved update frequency is limited by read granularity)

rec_attrs: 'true' | 'false' | 'force'
  Collect record attributes and include them in the spec with the
  data referencing it.
  - 'true':  collect data once at the beginning of first run;
  - 'false': turn off completely;
  - 'force': force (re)collection even if attributes have been already
             collected; useful if code has been changed in the system.

samples: pos_integer() | 'false' | 'infinity'
  Maximum number of samples to collect at each spec node.
  NB.: using 'infinity' is definitely not recommended, as the spec
  size will grow without bounds unless the number of processed
  records is limited to a small number!

term: 'dumb' | 'vt100'
  Terminal setting useful to override the $TERM environment variable;
  set it to 'dumb' to forcibly disable progress line rewrites.