Release 0.3

[Mec-iS]

This is the release candidate for v0.3

Please take a look, run some tests, etc.

Changes compared to development:

• improve documentation
• make default feature empty (no optional dependencies) refactor feature system
• cleanup unused variables

[codecov-commenter]

Codecov Report

❌ Patch coverage is 54.54545% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 43.97%. Comparing base (aab3817) to head (4558be5).
⚠️ Report is 83 commits behind head on development.

Files with missing lines	Patch %	Lines
src/tree/decision_tree_classifier.rs	50.00%	2 Missing ⚠️
src/tree/decision_tree_regressor.rs	50.00%	2 Missing ⚠️
src/cluster/kmeans.rs	0.00%	1 Missing ⚠️

Additional details and impacted files

@@               Coverage Diff               @@
##           development     #235      +/-   ##
===============================================
+ Coverage        43.93%   43.97%   +0.04%     
===============================================
  Files               85       85              
  Lines             7290     7281       -9     
===============================================
- Hits              3203     3202       -1     
+ Misses            4087     4079       -8     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Mec-iS]

@morenol please take a look

we have to fix the complications with the random numbers range generator at #214

we have now a clearer setup for the features:

• default: only base dependencies (no optional), should work with Wasi
• serde: whatever requires serialization
• datasets: whatever requires datasets
• std
• other

I don't understand the role of the std feature, I suppose is meant to use the standard library randomization library.
Now, we have to assure that:

• Wasm/Wasi (default) use the right implementation for random (currently getrandom),
• std feature uses the standard random
• datasets is currently using rand_distr, is there the possibility to make it use std random or getrandom?

[Mec-iS]

Fixed!

Features are now:

[features]
default = []
serde = ["dep:serde"]
ndarray-bindings = ["dep:ndarray"]
datasets = ["dep:rand_distr", "std_rand", "serde"]
std_rand = ["rand/std_rng", "rand/std"]

[morenol]

Let me check if there are issues with enabling js always.

[Mec-iS]

do we need it right?
I assumed js is just an additional feature, shouldn't affect the base functionality

[Mec-iS]

we need an alternative for this branch of the match

[morenol]

So, this feature flag only has effect for wasm32-unknown-unknown. Enabling this feature breakes the usage of wasm32-unknown-unknown with wasmtime and wasmer because it adds wasmbindgen to the build

[morenol]

We can avoid that branch by setting a seed for those targets. In any case I think that we could use get_random but only enabling "js" feature if it is a explicitly enabled feature

[morenol]

I mean, I think that we should keep this line in the features section:

js = ["getrandom/js"]

and remove js from being enabled by default in this line

[morenol]

my bad it seems to not be working with this change

[morenol]

Can we move all the changes related to rand vs getrandom to another PR? so we can merge the other changes?

[morenol]

@Mec-iS I think that we shouldnt merge this PR