Why am I seeing repeated `diff_url`s?

[jose]

Hi,

Let's take for example, the falcon Python project from the falconry organization and this diff_url. There are, in the BugSwarm dataset, 13 images corresponding to this diff_url, each with its own Job ID, Build ID, etc.:

falconry-falcon-30823226071
falconry-falcon-30823226301
falconry-falcon-30823227282
falconry-falcon-30823227361
falconry-falcon-30823227617
falconry-falcon-30823227692
falconry-falcon-30823227759
falconry-falcon-30823227836
falconry-falcon-30823227929
falconry-falcon-30823227988
falconry-falcon-30823228066
falconry-falcon-30823228169
falconry-falcon-30823228234

Overall, there are 2,073 unique diff_urls out of the 3,947 entries listed here.

My question is, should one consider this as (a) a single bug with multiple artifacts or (b) multiple bugs each with its own artifact? I would say the former...

[Robert-Furth]

Hi José,

The reason we consider those artifacts separate is because each was a separate GitHub Actions job, albeit grouped together in the same workflow run ("build" in BugSwarm's terminology). They have the same diff, but were run with different environment parameters; e.g. falconry-falcon-30823227282 was run under CPython 3.12, while falconry-falcon-30823227692 was run under PyPy 3.9. Since these differences can affect whether a job fails or passes, we keep them separate.

Checking the "Build ID -> Unique" checkbox in the sidebar on the website should group each artifact by their build/workflow run ID, but I just noticed that this doesn't seem to be working right. We'll get started on correcting that.