Consider project size in overall statistics

[bachp]

Currently all projects have an equal weight in the final metrics calculation.

So a project with 10 lines of code has the same weight as a 10'000 lines of code project.

[max-wittig]

Good point. Some things to consider.

• It's not a quick task to get lines of code of a project (runtime would be too long)
• Commits could be used as alternative statistic, but unsure if this number has much value
• Extend the GitLab API so that it doesn't just return the percentage.
  (Todo: Check how feasible that is)

[aram535]

You can get use '/projects/:id/statistics' to get the total_size of a project. It doesn't seem to be calculated on request so it's just a fetch of active data.

[max-wittig]

I guess, but what does the size have todo with the languages in the repository.

Ideally we would need the line count per language, which GitLab might consider to be an enterprise only feature 😦
https://gitlab.com/gitlab-org/gitlab-ce/issues/31010#note_187978381

[aram535]

I wasn't advocating the use of it -- as I think it's useless number for reporting purposes of large data set (~5000 projects). Was just pointing out that it's available - that's all.