Open DilumAluthge opened 3 years ago
We ignore it: https://github.com/JuliaEcosystem/PackageAnalyzer.jl/blob/a7c2b1d80fce25326fb4d679195cfb9edaec4f68/src/PackageAnalyzer.jl#L507-L508 :stuck_out_tongue:
It would be pretty easy to modify PackageAnalyzer to answer (1). All the "analysis" takes places in https://github.com/JuliaEcosystem/PackageAnalyzer.jl/blob/a7c2b1d80fce25326fb4d679195cfb9edaec4f68/src/PackageAnalyzer.jl#L480-L541. At that point of the pipeline, we've cloned the package to a local directory and are populating a Package
struct with information gained by inspecting the repo. So we could add a compat_helper
field to the struct and check if the workflow exists in that function, either by direct name or actually looping through workflow file contents and looking for the string "CompatHelper" or such. (I'd be in favor of such an addition since I think it's useful to know!)
For (2) and (3), I think that might be doable by combining the results of (1) with queries to the github api to check if an account is an org or not. I think that step might be outside the purview of PackageAnalyzer itself though.
I think that in the future we may want to collect all filenames in .github/workflows
, and exclude compathelper.yml
, tagbot.yml
only as part of data munging analysis, much like what we do now for the contributors (we were initially filtering out the bots, including @staticfloat
), but honestly I'd like to avoid changing the data structure before JuliaCon.
I'm interested in answering the following questions for the packages in the General registry.
For the purpose of this analysis, let us define
package X has CompatHelper installed
ason the default branch of the Git repo for package X, there is a file named .github/workflows/CompatHelper.yml
. (I'm open to a better definition.)Would it be possible for me to use PackageAnalyzer.jl to help me answer these questions?