hunting for bugs: history repeats

Almost whole day (today) I spent hunting for ralg() bug. It yields difference with OpenOpt for MATLAB (with my test case, where I noticed incorrect behaviour) only after 4th iteration. When I finally fix the one, I remembered that I had spent similar time for same bug in my OpenOpt for MATLAB (year ago) and several hours in OpenOpt for fortress (some months ago) (there are nothing but the ralg for now).