Why does this matlab function throw an error on the first call and how do I prevent the error

I am using the errorbar_tick function from https://www.mathworks.com/matlabcentral/fileexchange/22826-adjust-error-bar-width.

I have made some minor modifications to make it work with the 2016a version. This function throws an error the first time I call it, stating that: Index exceeds matrix dimensions. Error in errorbar_tick (line 48) x = h.Bar.VertexData(1,:); On a repeated call, it works. What is going on here and how can I fix this problem?

The modified code, in its entirety, is below.

function errorbar_tick(h,w,xtype)

%ERRORBAR_TICK Adjust the width of errorbars
%   ERRORBAR_TICK(H) adjust the width of error bars with handle H.
%      Error bars width is given as a ratio of X axis length (1/80).
%   ERRORBAR_TICK(H,W) adjust the width of error bars with handle H.
%      The input W is given as a ratio of X axis length (1/W). The result 
%      is independent of the x-axis units. A ratio between 20 and 80 is usually fine.
%   ERRORBAR_TICK(H,W,'UNITS') adjust the width of error bars with handle H.
%      The input W is given in the units of the current x-axis.
%
%   See also ERRORBAR
%

% Author: Arnaud Laurent
% Creation : Jan 29th 2009
% MATLAB version: R2007a
%
% Notes: This function was created from a post on the french forum :
% http://www.developpez.net/forums/f148/environnements-developpement/matlab/
% Author : Jerome Briot (Dut) 
%   http://www.mathworks.com/matlabcentral/newsreader/author/94805
%   http://www.developpez.net/forums/u125006/dut/
% It was further modified by Arnaud Laurent and Jerome Briot.

% Check numbers of arguments
narginchk(1,3)

% Check for the use of V6 flag ( even if it is depreciated ;) )
flagtype = get(h,'type');

% Check number of arguments and provide missing values
if nargin==1
  w = 80;
end

if nargin<3
   xtype = 'ratio';
end

% Calculate width of error bars
if ~strcmpi(xtype,'units')
    dx = diff(get(gca,'XLim'));  % Retrieve x limits from current axis
    w = dx/w;                   % Errorbar width
end

% Plot error bars
if strcmpi(flagtype,'errorbar') % ERRORBAR(...)
    x = h.Bar.VertexData(1,:); % Retrieve bar xdata from errorbar 
    dp = length(x)/3; % 3 data points per error bar 
    m = 1; % Multiplier for addition/subtraction 
    for ii = 1:dp 
        m = -1*m; % Switch between subtraction and addition 
        x(dp+ii:dp:end) = x(ii)+m*w/2; % Change xdata with respect to the chosen ratio 
    end 
    h.Bar.VertexData(1,1:end) = x;
  else
      error('Please enter an ErrorBar object!');
end

[John D'Errico]

Why not use the debugger? After all, you were the one who made the changes, so you know what you did.

Answer 1

[Joshua Knicely]

The answer is in the comments of this question: https://www.mathworks.com/matlabcentral/answers/264534-how-can-i-change-the-width-of-horizontal-lines-of-error-bars-in-errorbar-plot-in-matlab-2015b

According to Lamia Kasmi:

" Thanks, indeed it works in the command window, however it doesn't in a script. The solution to that is that one needs to add a "drawnow" after calling the errorbar function and before calling the errorbar_tick function, otherwise errorbar_tick might be faster than the plotting and it might look for the handle before it is even created by the errorbar function. So in summary, in a script, one should use errorbar_tick in the following way:

Example:

x = [1 2 3 4];

y = [1 2 1 2];

dy = [0.1 0.1 0.1 0.1];

h = errorbar(x,y,dy);

drawnow

errorbar_tick(h) "