src/glsl/TODO

   1 - Handle currently unsupported constant expression types
   2   - ir_unop_sign
   3   - ir_unop_exp2
   4   - ir_unop_log2
   5   - ir_unop_u2f
   6   - ir_unop_trunc
   7   - ir_unop_ceil
   8   - ir_unop_floor
   9   - ir_unop_sin
  10   - ir_unop_cos
  11   - ir_binop_min
  12   - ir_binop_max
  13   - ir_binop_pow
  14
  15 - Handle constant expressions of (struct == struct)
  16
  17 - Handle constant expressions of (struct != struct)
  18
  19 - Add support to ir_constant for array constants Arrays can only be
  20   - declared 'const' in GLSL 1.20+.  This is because there are no
  21     array constructors in GLSL 1.10, and any variable declared as
  22     'const' must have an initializer.
  23
  24 - Handle constant expressions of (array == array)
  25
  26 - Handle constant expressions of (array != array)
  27
  28 - Treat built-in functions with constant parameters as constant expressions.
  29   - Rewrite all built-in functions return a single expression.
  30   - Modify the HIR generator for functions to automatically inline built-in
  31     functions durning translation.
  32   - Care must be taken to handle both the 1.10 rules and the 1.20+ rules.  In
  33     1.10, built-in functions cannot be constant expressions.
  34
  35 - Detect code paths in non-void functions that don't reach a return statement
  36
  37 - Handle over-riding built-in functions
  38   - Is the overload per-compilation unit or per-linked shader?
  39
  40 - Handle redeclaration of built-in variables
  41   - Handle addition of qualifiers such as 'invariant' or 'centroid'.
  42   - Handle resizing of arrays.
  43   - Other?  We'll have to look at the spec.
  44
  45 - Improve handling of constants and their initializers.  Constant initializers
  46   should never generate any code.  This is trival for scalar constants.  It is
  47   also trivial for arrays, matrices, and vectors that are accessed with
  48   constant index values.  For others it is more complicated.  Perhaps these
  49   cases should be silently converted to uniforms?
  50
  51 1.30 features:
  52
  53 - Implement AST-to-HIR conversion of bit-shift operators.
  54
  55 - Implement AST-to-HIR conversion of bit-wise {&,|,^,!} operators.
  56
  57 - Implement AST-to-HIR conversion of switch-statements
  58   - switch
  59   - case
  60   - Update break to correcly handle mixed nexting of switch-statements
  61     and loops.
  62
  63 - Handle currently unsupported constant expression types
  64   - ir_unop_bit_not
  65   - ir_binop_mod
  66   - ir_binop_lshift
  67   - ir_binop_rshift
  68   - ir_binop_bit_and
  69   - ir_binop_bit_xor
  70   - ir_binop_bit_or
  71
  72 - Implement support for 1.30 style shadow compares which only return a float
  73   instead of a vec4.