Propagate static(0) * x

[Tokazama]

Reverts to prior behavior and fixes #58. @chriselrod, It seems like we should only return StaicInt{0} when multiplied with other integers because we should promote. Would that break anything on your end?

[codecov]

Codecov Report

Merging #59 (31b6e62) into master (16f441f) will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #59      +/-   ##
==========================================
+ Coverage   98.80%   98.82%   +0.01%     
==========================================
  Files           8        8              
  Lines         417      424       +7     
==========================================
+ Hits          412      419       +7     
  Misses          5        5              

Impacted Files	Coverage Δ
src/float.jl	98.14% <100.00%> (+0.07%)	⬆️
src/int.jl	100.00% <100.00%> (ø)

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[chriselrod]

Taking my example from JuliaArrays/ArrayInterface.jl#262 (comment)
On this PR, I get

julia> using VectorizationBase

julia> using Static: Zero

julia> vx = Vec(ntuple(_->randn(),pick_vector_width(Float64))...)
Vec{8, Float64}<0.11215394134606615, -1.1365862303670673, 0.7231326012882932, -1.280896639969639, -0.41616032433006667, -0.2873581365771947, 0.3625756468339046, 0.7963550919238662>

julia> vx * Zero()
static(0)

julia> Zero() * vx
static(0)

[chriselrod]

It seems like we should only return StaicInt{0} when multiplied with other integers because we should promote. Would that break anything on your end?

The failing VectorizationBase test wants to preserve static(0), even with Vec{W,Float64}. I'd prefer that behavior to be consistent with Float64.

[Tokazama]

It seems like we should only return StaicInt{0} when multiplied with other integers because we should promote. Would that break anything on your end?

The failing VectorizationBase test wants to preserve static(0), even with Vec{W,Float64}. I'd prefer that behavior to be consistent with Float64.

So merge this as is?

[chriselrod]

Yes