[Bug-glpk] Re: LPX_FX is a special case of LPX_DB

[Andrew Makhorin]

>If the user supplies a row/column of type LPX_DB with lb == ub,
>internally set its type as LPX_FX.  After all, according to the
>former type, lb <= x <= ub, which implies that lb == ub is valid.
>But, currently, this creates an error when solving the model.

For some reasons it was decided not to change any data supplied by the
user. Besides, it would be not clear how to understand the case when lb
and ub are close to each other. I think it is not burdensome to write
something like follows:

   lpx_set_col_bnds(lp, lb == ub ? LPX_FX : LPX_DB, lb, ub);

or

   lpx_set_col_bnds(lp, fabs(lb - ub) <= 1e-6 ? LPX_FX : LPX_DB, lb, ub);