[OT] Leverage Points

[Mike Franklin]

On Saturday, 18 August 2018 at 13:33:43 UTC, Andrei Alexandrescu 
wrote:

> where are the best leverage points in making the D language 
> more successful.

I'm still internalizing the article and thinking about how it 
applies to the "D system", but I've always thought facilitating 
the incorporation of GDC into GCC to be the single most 
accelerating thing we could do to gain more adoption.  It 
somewhat fits into *7. The gain around driving positive feedback 
loops*.

But there's risk associated with that.  Walter has often said 
that "build it and they will come" is a Hollywood myth, but I 
disagree.  Part of the reason why D hasn't achieved mass 
adoption, isn't because it's not marketed well, but because it 
has a number of flaws.  Most of us see the *potential* of D, and 
are able to look past the flaws, with the faith (hopefully not 
misplaced) that they will one day be addressed.  Others only see 
the flaws and the appeal of other programming languages with more 
resources, better management, more talent, and especially more 
velocity toward their goals.

I often worry that if we encourage adoption, before we have 
something worthy of adoption, we'll only leave users with a bad 
taste in their mouth [0].  I've already seen a number of people, 
some major contributors, leave D for greener pastures.  Most of 
the contributors that built the D runtime and did the majority of 
bug fixing in the compiler are gone now.  At this point in time, 
I can only recommend D professionally to teams that are risk 
takers, have the aptitude to solve their own problems, and have 
the resources and willingness to be D contributors.

We should probably be looking more for leverage points to help us 
better capitalize on the resources and talent we have and bring 
in more.  Unfortunately I'm seeing an over-correction in *8. The 
strength of negative feedback loops, relative to the impacts they 
are trying to correct against*.  As we try to get contributors to 
focus on the things that matter (at least to the powers that be), 
we frustrate them until they close their pull requests or just 
give up [1] [2].

It took me a few years to find my "in", and I'm still not really 
"in", but I learned that the *little things* that some consider a 
distraction are how people get started contributing to D.  I've 
often said that we actually don't need more contributors; but 
more reviewers.  There's a catch to that, though; they're not 
going to become reviewers if they can't first become 
contributors.  So perhaps, I need to correct my perspective.

So, I'll close with this:  We should probably be more welcoming 
to those willing to contribute, let them work on the little stuff 
that's important to them, throw them a bone or two, review their 
pull requests in a timely manner, etc... I think those 
contributors will eventually become our reviewers, and then they 
will eventually lessen the burden so veterans can focus on the 
things that they think are higher priorities.  This is a positive 
feedback loop.  Help people become positive contributors, and 
those contributors will eventually help the next generation.  I 
think there are a few little things the leadership, especially, 
can do to prime that pump, starting with being more active, 
helpful, and gracious with things that are currently sitting in 
the PR queue.  Though it's a two-way street, and some 
contributors could also be more cooperative also.

Walter and a few others have been quite gracious to me [3] [4].  
I've tried to pay that forward and help other contributors find 
their "in", but I'm still not able to review and make decisions 
about many things, so I'm only of limited help.  I don't think 
others have been treated as well.

Mike

[0] - https://issues.dlang.org/show_bug.cgi?id=14100 - Link in 
that issue no longer exists, but let's just say the user wasn't 
happy with D
[1] - 
https://github.com/dlang/dmd/pulls?q=is%3Apr+author%3Amarler8997+is%3Aclosed
[2] - https://github.com/dlang/dmd/pull/8378
[3] - 
https://github.com/dlang/dmd/pull/7395#issuecomment-349200847
[4] - 
https://github.com/dlang/dmd/pull/7055#issuecomment-320006283