On Saturday, 30 December 2017 at 14:42:45 UTC, Muld wrote:
On Saturday, 30 December 2017 at 06:55:13 UTC, Walter Bright
wrote:
It's not like we have a shortage of bugzilla issues and are
wondering what to do next.
Yah there are a ton of Bugzilla issues, that's the problem.
More than half of them aren't "actionable" as you put it.
Here's the problem, look at something like Rust:
Pull requests? 95 open, it's about the same as Dlang, But if
you go to the last page...
https://github.com/rust-lang/rust/pulls?page=4&q=is%3Apr+is%3Aopen
Look at that the oldest one is from October 15th, 20_17_.
Now we go to DMD...
https://github.com/dlang/dmd/pulls?page=6&q=is%3Apr+is%3Aopen
Oldest one is from January 17, 20_13_.
This is a problem that many of us are working on fixing. The main
reason many of these old zombie PRs stick around is that
historically, people are hesitant to close things (for a variety
of social reasons, I feel). While there is still the slightest
chance that something might someday be merged, it is kept open.
Rust is a lot more aggressive about closing bad or outdated PRs
and either guiding PRs that need work to get to a mergeable
state, or closing them and communicating that this is not the
correct way to go. I watched a talk by a Rust contributor
specifically on this point awhile ago - they have a bot that does
a lot of the PR closure work to get around the fact that people
are hesitant to be the "bad guy" and tell someone that their work
is not good enough. D needs to get much better at this, and I
think things are happening - slowly. The bad optics and
demoralizing effect of letting things sit forever without
definitive action outweighs the potential loss from being more
aggressive about closing or merging.