PyDataBlog / ParallelKMeans.jl

Parallel & lightning fast implementation of available classic and contemporary variants of the KMeans clustering algorithm
MIT License
50 stars 13 forks source link

Extend compat: StatsBase = "0.32, 0.33, 0.34" #119

Closed ablaom closed 1 year ago

ablaom commented 1 year ago

@PyDataBlog

edit This PR also dumps support for Julia < 1.6, to get tests passing with minimal fuss.

codecov-commenter commented 1 year ago

Codecov Report

Merging #119 (176fb8f) into master (a542d02) will increase coverage by 0.03%. The diff coverage is n/a.

:exclamation: Current head 176fb8f differs from pull request most recent head 0d4f6c1. Consider uploading reports for the commit 0d4f6c1 to get more accurate results

@@            Coverage Diff             @@
##           master     #119      +/-   ##
==========================================
+ Coverage   98.34%   98.38%   +0.03%     
==========================================
  Files          10       10              
  Lines         909      992      +83     
==========================================
+ Hits          894      976      +82     
- Misses         15       16       +1     

see 9 files with indirect coverage changes

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

PyDataBlog commented 1 year ago

@Arkoniak everything working as expected. Benchmarks actions keep failing for some reason. Any ideas?

ablaom commented 1 year ago

working on it now

ablaom commented 1 year ago

Sorry, I don't know what the issue is. There is a package conflict that suggests that the master version of ParallelKMeans is being used in the benchmarking instead of the source branch, but I don't understand why. Locally I can get around this (on 1.8) but then then there is yet another pkg-environment issue with BenchmarkCI.judge(), which seems to switch to a new environment and then complains it can't find "PkgBenchmark".

I've no experience with this tool, so I think you'll need to find a fix or disable the benchmarking for now.

Arkoniak commented 1 year ago

This error is ok, just a small issue on BenchmarkCI which hard to overcome. What it does, is trying to run previous and current version of the package. But it creates the same environment for both of them, that's why one version can run, but another can not (it can't compile "official" 1.0.1version of ParallelKMeans in the environment where StatsBase v.0.34 is installed.

So, we can just accept this commit and next commit will run against proper version with proper environment without any errors.

ablaom commented 1 year ago

Okay. So can we merge this and get a new release please?

PyDataBlog commented 1 year ago

Okay. So can we merge this and get a new release, please?

I will bump up and release.

Arkoniak commented 1 year ago

Okay. So can we merge this and get a new release please?

Version 1.1.0 with required changes was released. Thank you for your PR.

ablaom commented 1 year ago

❤️