Modification of fminunc

It may not be a bug but something I am missing. In some cases fminuc starts a line search during the first iteration and returns the message that it cannot find a good solution along the search direction, although it went through some points with a lower cost (easily seen by printing the cost function at each call of the optimization function).
To solve that problem I modified one of the sub-function called by fminuc (fminusub) so that it checks that the best solution along the line search is not lower than the last best solution before claiming that it cannot find a good solution.
Well, I don?t know it is clear, but it is an easy fix. Of course I welcome any suggestion if someone has any experience with this issue.
Just replace the function fminsub.m in the folder: \toolbox\optim\private