BUGLIST

*********** THIS IS THE TO-DO LIST


**** BUGS PROPER

none known

**** OMISSIONS, CONCERNS, TO-DO

- preprocessor features almost inexistant.
- solver doesn't work with preprocessor turned on.


**** MISSING FEATURES

- Forrest-Tomlin update could be replaced by Suhl-Suhl or any other suitable
  update method. Concerns are different than in case of floating point implementations
  as we care little about numerical stability *directly* but only want to reduce
  numerical fill-in in U and in results during f- and b-trans as much as possible.
  A scheme with permutations (c.f. Saunders, Reid...) should be possible.

- A cost perturbation scheme should be implemented to avoid dual degenaracy, 
  and hence, a primal simplex algorithm is necessary

- At least some basic parallelism, many computations can (and should) be offloaded
  into another thread. Off the top of my head: DSE Ftran computation, Basis matrix
  refactorization or update, Primal values update after bound flips...