function ft_realtime_selectiveaverage(cfg)
 
% FT_REALTIME_SELECTIVEAVERAGE is an example realtime application for online
% averaging of the data. It should work both for EEG and MEG.
%
% Use as
%   ft_realtime_selectiveaverage(cfg)
% with the following configuration options
%   cfg.channel    = cell-array, see FT_CHANNELSELECTION (default = 'all')
%   cfg.trialfun   = string with the trial function
%
% The source of the data is configured as
%   cfg.dataset       = string
% or alternatively to obtain more low-level control as
%   cfg.datafile      = string
%   cfg.headerfile    = string
%   cfg.eventfile     = string
%   cfg.dataformat    = string, default is determined automatic
%   cfg.headerformat  = string, default is determined automatic
%   cfg.eventformat   = string, default is determined automatic
%
% To stop the realtime function, you have to press Ctrl-C
 
% Copyright (C) 2008, Robert Oostenveld
%
% Subversion does not use the Log keyword, use 'svn log <filename>' or 'svn -v log | less' to get detailled information
 
% set the default configuration options
if ~isfield(cfg, 'dataformat'),     cfg.dataformat = [];      end % default is detected automatically
if ~isfield(cfg, 'headerformat'),   cfg.headerformat = [];    end % default is detected automatically
if ~isfield(cfg, 'eventformat'),    cfg.eventformat = [];     end % default is detected automatically
if ~isfield(cfg, 'channel'),        cfg.channel = 'all';      end
if ~isfield(cfg, 'bufferdata'),     cfg.bufferdata = 'last';  end % first or last
 
% translate dataset into datafile+headerfile
cfg = ft_checkconfig(cfg, 'dataset2files', 'yes');
cfg = ft_checkconfig(cfg, 'required', {'datafile' 'headerfile'});
 
% ensure that the persistent variables related to caching are cleared
clear read_header
% start by reading the header from the realtime buffer
hdr = ft_read_header(cfg.headerfile, 'cache', true);
 
% define a subset of channels for reading
cfg.channel = channelselection(cfg.channel, hdr.label);
chanindx    = match_str(hdr.label, cfg.channel);
nchan       = length(chanindx);
if nchan==0
  error('no channels were selected');
end
 
prevSample = 0;
count      = 0;
 
% initialize the timelock cell-array, each cell will hold the average in one condition
timelock = {};
 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% this is the general BCI loop where realtime incoming data is handled
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
while true
 
  % determine latest header and event information
  event     = read_event(cfg.dataset, 'minsample', prevSample+1);  % only consider events that are later than the data processed sofar
  hdr       = read_header(cfg.dataset, 'cache', true);             % the trialfun might want to use this, but it is not required
  cfg.event = event;                                               % store it in the configuration, so that it can be passed on to the trialfun
  cfg.hdr   = hdr;                                                 % store it in the configuration, so that it can be passed on to the trialfun
 
  % evaluate the trialfun, note that the trialfun should not re-read the events and header
  fprintf('evaluating ''%s'' based on %d events\n', cfg.trialfun, length(event));
  trl = feval(cfg.trialfun, cfg);
 
  % the code below assumes that the 4th column of the trl matrix contains the condition index
  % set the default condition to one if no condition index was given
  if size(trl,2)<4
    trl(:,4) = 1;
  end
 
  fprintf('processing %d trials\n', size(trl,1));
 
  for trllop=1:size(trl,1)
 
    begsample = trl(trllop,1);
    endsample = trl(trllop,2);
    offset    = trl(trllop,3);
    condition = trl(trllop,4);
 
    % remember up to where the data was read
    prevSample  = endsample;
    count       = count + 1;
    fprintf('processing segment %d from sample %d to %d, condition = %d\n', count, begsample, endsample, condition);
 
    % read data segment from buffer
    dat = ft_read_data(cfg.datafile, 'header', hdr, 'begsample', begsample, 'endsample', endsample, 'chanindx', chanindx, 'checkboundary', false);
 
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    % from here onward it is specific to the processing of the data
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
    % put the data in a fieldtrip-like raw structure
    data.trial{1} = dat;
    data.time{1}  = offset2time(offset, hdr.Fs, endsample-begsample+1);
    data.label    = hdr.label(chanindx);
    data.hdr      = hdr;
    data.fsample  = hdr.Fs;
 
    % apply some preprocessing options
    data.trial{1} = preproc_baselinecorrect(data.trial{1});
 
    if length(timelock)<condition || isempty(timelock{condition})
      % this is the first occurence of this condition, initialize an empty timelock structure
      timelock{condition}.label   = data.label;
      timelock{condition}.time    = data.time{1};
      timelock{condition}.avg     = [];
      timelock{condition}.var     = [];
      timelock{condition}.dimord  = 'chan_time';
      nchans   = size(data.trial{1}, 1);
      nsamples = size(data.trial{1}, 2);
      % the following elements are for the cumulative computation
      timelock{condition}.n   = 0;                          % number of trials
      timelock{condition}.s   = zeros(nchans, nsamples);    % sum
      timelock{condition}.ss  = zeros(nchans, nsamples);    % sum of squares
    end
 
    % add the new data to the accumulated data
    timelock{condition}.n  = timelock{condition}.n  + 1;
    timelock{condition}.s  = timelock{condition}.s  + data.trial{1};
    timelock{condition}.ss = timelock{condition}.ss + data.trial{1}.^2;
 
    % compute the average and variance on the fly
    timelock{condition}.avg = timelock{condition}.s ./ timelock{condition}.n;
    timelock{condition}.var = (timelock{condition}.ss - (timelock{condition}.s.^2)./timelock{condition}.n) ./ (timelock{condition}.n-1);
 
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    % from here onward additional processing of the selective averages could be done
    % as an example here the ERP of each condition is plotted in its own figure
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
    % compute the t-score versus zero by dividing the average by the standard error of mean
    tscore = timelock{condition}.avg ./ (sqrt(timelock{condition}.var)./(timelock{condition}.n - 1));
 
    figure(condition)
    plot(timelock{condition}.time, tscore);
    title(sprintf('condition %d, ntrials = %d', condition, timelock{condition}.n));
 
    % force matlab to redraw the figure
    drawnow
 
  end % looping over new trials
end % while true