datetick, reason for explicitly undocumented behavior?

[Nicholas Jankowski]

this came up in a StackExchange query today [1]. A user had transposed inputs to the datetick command

datetick(29, 'x') versus datetick('x', 29)

While this obviously failed to produce the desired output, the command didn't error out as it it does in Matlab.  Instead, it instead followed an undocumented codepath that allows you to specify the starting date to use for the labels. While it's not overly odd that Octave code have options that extend beyond Matlab's, I find it odd that it's not only undocumented but explicitly undocumented [2]:

  ## Don't publish the existence of this variable for use with dateaxis
  if (length (varargin) > 0)
    startdate = varargin{1};
  else
    startdate = [];
  endif

What would be the reason for not documenting a known and deliberate extension from 'standard' m-code?

nickj

[1] http://stackoverflow.com/questions/41376987/unexpected-behavior-of-datetick-in-octave
[2] http://hg.savannah.gnu.org/hgweb/octave/file/57e82b74f89b/scripts/plot/appearance/datetick.m#l138