Open GitHub issues. That is, all GitHub issues including bug,
documentation, enhancement, performance, throughput, etc.
and excluding only cxx20, cxx23, and LWG to avoid double-counting.
Skipped tests in the libcxx test suite. There are many reasons for skipped tests:
STL bugs, test bugs, compiler bugs, missing STL features, missing compiler features, etc.
To avoid double-counting, this chart excludes skipped tests due to missing STL features.
We don't count "patch papers" as extra work when one paper adds a feature
and another paper modifies that feature before we've implemented it.
This chart weights features equally, instead of weighting
by usefulness for users or difficulty for implementers.
The average age of open pull requests, in days.
The average "waiting time" of open pull requests, in days. This measures how long it's been since a
maintainer submitted a review requesting changes. Any other activity doesn't reset this clock.
The combined age of open pull requests. The unit is PR-months (like kilowatt-hours);
for example, 5 PRs that were all opened 2 months ago have a combined age of 10 PR-months.
The combined "waiting time" of open pull requests, in PR-months.
For this chart, more is better. It includes merges into any branch
and excludes PRs that were closed without being merged, which are rare.
Line: How many PRs have been merged over the past month.
This uses a "30 day" sliding window, with some smoothing:
PRs merged 0 to 20 days ago are weighted at 100%.
PRs merged 20 to 40 days ago are weighted from 100% to 0%.
Bars: How many PRs were merged in each calendar month.