alan-turing-institute / PDSampler.jl

Piecewise Deterministic Sampler library (Bouncy particle sampler, Zig Zag sampler, ...)
Other
33 stars 8 forks source link

Julia7plus #34

Closed tlienart closed 5 years ago

tlienart commented 5 years ago

This branch brings 0.7 and 1.0 support with fixes all over to match the introduced breaking changes. Will merge this unless it doesn't pass on Travis for some reason.

The past release will serve as legacy for 0.6 should there be any users willing to use that.

This branch does not introduce new features or breaking API changes.

tlienart commented 5 years ago

Ok so it passes on nightly (1.0) by the looks of it but fails on 0.7.

most likely this is because I'm using operations on generators which are fine with 1.0 but not 0.7.

abs.(e for e in [1, -2, 3])

this works on 1.0 but not on 0.7. However adding a collect works on both

abs.(collect(e for e in [1, -2, 3]))

so I'll just hunt for those (they appear in the tests) and fix that.

tlienart commented 5 years ago

Ok, it passes on 0.7 unix so it's going to pass on everything else, I'll merge as soon as that's all done.

tlienart commented 5 years ago

yeah that's all good, AppVeyor does something silly on windows, not being able to find the stdlib LinearAlgebra, I'm not going to bother and will just remove AppVeyor since it's not like it's adding much info to the travis build anyway.

codecov-io commented 5 years ago

Codecov Report

Merging #34 into master will decrease coverage by 1.16%. The diff coverage is 100%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master      #34      +/-   ##
==========================================
- Coverage   97.23%   96.06%   -1.17%     
==========================================
  Files          12       12              
  Lines         506      407      -99     
==========================================
- Hits          492      391     -101     
- Misses         14       16       +2
Impacted Files Coverage Δ
src/simulate.jl 81.35% <ø> (-3.46%) :arrow_down:
src/local/simulate.jl 100% <100%> (ø) :arrow_up:
src/local/factorgraph.jl 100% <100%> (ø) :arrow_up:
src/geometry.jl 100% <100%> (ø) :arrow_up:
src/ippsampler.jl 90.9% <100%> (-5.82%) :arrow_down:
src/local/event.jl 100% <100%> (ø) :arrow_up:
src/models/mvgaussian.jl 100% <100%> (ø) :arrow_up:
src/path.jl 100% <100%> (ø) :arrow_up:
src/models/tmvgaussian.jl 100% <100%> (ø) :arrow_up:
... and 8 more

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 17e3108...704ec25. Read the comment docs.

tlienart commented 5 years ago

Ha no actually the appveyor script was just wrong, it was still the 0.6 version, fixed that with the new appveyor script. Should now be all good to merge.