fix evaluating one/zero in generator body

[KristofferC]

This should fix #7 cc @andyferris

[codecov-io]

Codecov Report

Merging #8 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master      #8      +/-   ##
=========================================
+ Coverage   96.59%   96.6%   +0.01%     
=========================================
  Files          13      13              
  Lines         675     677       +2     
=========================================
+ Hits          652     654       +2     
  Misses         23      23

Impacted Files	Coverage Δ
src/constructors.jl	98.14% <100%> (+0.07%)	✅

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 49d600a...1454860. Read the comment docs.

[KristofferC]

Heh.. was going to put you on review and you did it yourself 17 sec ago... Same wavelength brooo.

[andyferris]

I didn't look over the all the codebase - is this the only place that this kind of thing happens? Does it solve the using vs include error you saw?

[fredrikekre]

Does it solve the using vs include error you saw?

No.

[KristofferC]

Problem is here https://github.com/KristofferC/Tensors.jl/blob/49d600a05c07407f410bdf74ebe9eae444748929/src/constructors.jl#L42

I guess it is the same if the type is defined after that function (get_type).

[andyferris]

Yeah, I saw that and it definitely raised some red flags in my mind, but I didn't exactly know what they were. It seemed a bit like the similar_type problem in StaticArrays (similar_type is the counterpart to similar for immutable arrays).

[KristofferC]

But it is weird because the Tensor type is defined at that point. We don't expect anyone to subtype these types so we should have it easier than StaticArrays. Anyway, I'm not sure what the difference is between first include the module file and then using it ... Perhaps the @generated stuff runs but the types are still not really defined... I dunno. :/

[andyferris]

Unlike most of Julia code, the order of generated functions w.r.t. other method definitions is important, even if they are simultaneously loaded from the same file. Maybe shuffling the order of stuff could help.

Not sure why using vs include should cause any difference there, however... Seems a mystery to me.

[fredrikekre]

Maybe shuffling the order of stuff could help.

Ref #9

Thanks for the tip!

[andyferris]

No worries. :)