Closed joergbuchwald closed 3 years ago
do you have an idea how to fix the pipeline? The output is nondeterministic. Does it make sense in your opinion to remove the no-deterministic part in the show overload?
I use a seed at the documentation makefile in order to get deterministic results. Seeded tests are standard practice, and we can use a set of seeds to try more variation.
You can build docs locally to get the proper output by doing:
$ cd ExperimentalDesign
$ julia
Then:
(@v1.6) pkg> activate docs
Activating environment at `~/.julia/dev/ExperimentalDesign/docs/Project.toml`
julia> using Documenter, ExperimentalDesign
julia> random_seed = 443591
julia> DocMeta.setdocmeta!(ExperimentalDesign,
:DocTestSetup,
:(using ExperimentalDesign, Distributions,
Random, StatsModels, DataFrames;
Random.seed!($random_seed););
recursive = true)
These are the steps on the docs makefile, and will allow you to test the current version:
julia> doctest(ExperimentalDesign)
To update the doctests with this seed, run:
julia> doctest(ExperimentalDesign; fix = true)
This will change your local files, so make sure all other tests pass and commit the changes.
add a Latin Hypercube sampling design constructed by LatinHypercubeSampling.jl