Input error handling

[kuol]

Matrix Mode:

Input O:

• [ ] Missing required columns (variable, coefficient)
• [ ] Make certain columns optional (lb, ub, type) — set default values:
  1. lb = 0
  2. ub = Inf
  3. type = C
• [ ] Handle wrong letter case
• [ ] Handle missing column names (but the number of columns and their order is correct)
• [ ] Column type:
  1. coefficient, lb, ub: numbers (strings that can be cast to numbers)
  2. type: must be “C”, “B” or “I”

     Input A:

Slam matrix:

• [ ] Missing required columns (i, j, v)
• [ ] Handle wrong letter case
• [ ] Handle missing column names (but the number of columns and their order is correct)
• [ ] Column type: all must be numbers (strings that can be cast to numbers)

Dense matrix:

• [ ] Check if number of columns = number of rows in Input O
• [ ] Column type: all must be numbers (strings that can be cast to numbers)

  Input B

This is an optional input, there is an option to combine integrate Input B to Input A:

• [ ] Missing required columns (i, j, v)
• [ ] Handle wrong letter case
• [ ] Handle missing column names (but the number of columns and their order is correct)
• [ ] Column type:
  1. dir: only allows “<=”, “>=”, “=” —we also need to handle change “==” to “=”, in case some users accidentally use “==”
  2. rhs: only allows numbers (string that can be cast to numbers)

[ramnathv]

Some more notes on this issue. It repeats some of what we already have. We need to consolidate it.

O input

1. should always have two fields named variable and coefficient.
2. can additionally have column named
   • type, which must take values in C, I, B.
   • lb, to indicate lower bound.
   • ub to indicate upper bound.
   • description for a more detailed description that appears in the report.
3. user can optionally add any number of additional columns with information.

A input

1. any columns other than constraint, name, dir, rhs, description will be considered the coefficient matrix.
2. any rows named dir, rhs or description in the variable column will be treated specially.
3. logic for recognizing type of matrix:
   • if there are only three columns named i, j and v, then slam, constraints in rows.
   • if there is a column named variable, then it is dense_variable
   • otherwise, it will be interpreted as dense_constraints

Rules:

Lining up Variables

1. If A matrix does not have a column named variable or constraint, then the columns corresponding to the constraint matrix will be assumed to be in the same order as the variables in the O matrix.
2. Dense_Contraint -> Column names MUST be identical to those in the O matrix
3. Dense_Variable <- Column named variable should have identical values to the column named variable in the O matrix.

Lining Constraints

1. Constraints are assumed to be in the same order as the A matrix.