First my thanks to Authors for creating this nice program. It is not only useful program, it is also good to be used for those who study algorithms.
Now Feature suggestion:
Suppose I run a multiple generation of timetables, let say 10 time timetables are generated.
I understand each generation needs different random seeds.
From the 10 generated time tables, I may pick 1 or 2 which have low conflict percentage for analysis, and then again I may choose to generate with some slight data adjustments like teacher name change or others. When I re-generate again, I may not get the same time table because random seeds are different. If we could retain random seeds separately in a history (of generation) file, we can re-use those seeds and generate again.
Even though we can enter seeds by ourselves, if they are taken from the multiple-time table generations, we can pick the better seeds.
Thank you!
Your suggestion is already included.
The seed is always saved in the fet result directory "FILENAME_random_seed_before.txt".
Come to think of it, your idea "pick the better seed" is unneeded if you only changed the teacher name.
You can take the result (again in fet result directory "FILENAME_data_and_timetable.fet"), change the teacher name and genarate again. It doesn't matter how difficult your dataset is, regeneratng will take max a few seconds, because the result is fixed.