some symbolic instructions are failing

[zapashcanon]

see:

#96 (comment)
#96 (comment)
#96 (comment)

[epatrizio]

After #151 PR, instructions have been implemented.
However, it remains (non-exhaustive, in actual scope) :

• f32.copysign

  owi/test/sym/binop_f32.wat

  Line 39 in 4e3a35f

  ;; (f32.eq (f32.copysign (local.get $x) (f32.const 42)) (f32.const 0))

• f64.copysign

  owi/test/sym/binop_f64.wat

  Line 39 in 4e3a35f

  ;; (f64.eq (f64.copysign (local.get $x) (f64.const 42)) (f64.const 0))

• f32.trunc

  owi/test/sym/unop_f32.wat

  Line 35 in 4e3a35f

  ;; (f32.eq (f32.trunc (local.get $f32)) (f32.const 123))

• f64.trunc

  owi/test/sym/unop_f64.wat

  Line 35 in 4e3a35f

  ;; (f64.eq (f64.trunc (local.get $f64)) (f64.const 123))

[zapashcanon]

cc @filipeom, I believe these are the last (hopefully) missing instructions in encoding 😅

[filipeom]

I think I added trunc, copysign can probably be implemented without an encoding operator. I can take a look at these when I get more free time 😃

[zapashcanon]

@epatrizio could you do a PR where you re-enable the trunc tests then ? :)

[zapashcanon]

Thanks! Do you want to try implementing copysign in Owi directly too ? This should not be too hard.

[zapashcanon]

This has actually been fixed, thanks @epatrizio ! :)

@filipeom, for now copy_sign is implemented in Owi, if you think it would make sense to have it directly in encoding, feel free to copy-paste the implementation (and to patch Owi to use the one from encoding :-)).