It was some technical feat (PHP 5.2 compatibility? *cry*) and making architectural changes to plugin with multiple millions of users can give even me jitters.
But curiously quite a few people had since asked not about technical side (just go read case study I also posted), but people side.
How well did it go?
It is happening
For starters it is important to distinguish experimenting with a technology from implementing a technology. Experiments are a bet. They are hopes that something will bring improvement, usually based on second hand info and gut feeling.
Implementations are nothing like that. They bring in specific solution to a specific need. Usually long overdue need. Do or die.
Experiments are like making a snowman. Implementations are like riding an avalanche.
Be right person
When you are invited in for a very specific implementation it's probably not out of the blue.
Am I the most qualified person to work with Composer in general?
Not by far.
Had I created the site about Composer in WordPress, talked about Composer with WordPress at large WordPress conference, and generally nagged everyone around about Composer and WordPress for many months? Is “Composer all the things!” my personal catch phrase?
Yes to all of the above.
Have right blessing
And finally, it matters not as much why do I think I should be there, as why the person putting me there thinks so.
Boss said so. In a good place that's a trustworthy reason enough.
Bring a pillow
When you are working with something near and dear to your heart, it is easy to take everything personally. Come into every conversation with a sword and swing it at every perceived offense or opposition.
Step on it. Throw a sword out. Bring a pillow.
Your function for the period is to take every “attack” like a fluffy happy cloud would, then emit sunshine until any hint of negativity dries away.
You are available to everyone, any time of day and night, to resolve any tiniest question, concern, and worry they might possible have. Those they don't even know they have too.
There is no partial implementation. Did I say do or die?
The progress is gradual, but switch over must be as close to flawless as possible. There is no should have worked, anything less than faster and better is missing a point why are you there in first place.
Have a road map, complete a road map, flip a switch.
There are two kinds of automation — one runs inside our computers, another inside our heads. We get used to things and workflows.
When they change — it can really bite. Triple check that computers are doing as much as possible, and that people are really on a new page about the rest.
You are now Google
When switch is finally flipped the questions start. Imagine the Google is taking couple weeks off and for this specific thing you are the replacement.
Have all the answers. Well, not possible. Have most of the answers and be very ready to find the rest or make a plan of what it would take.
If people aren't literally saying “that was fast” then you are not answering fast enough.
And when it's done it's not done at all. It is now living and breathing part of the project, opening up new possibilities and improvements.
That was kind of the point, right? Get to it. :)
Having typed this out I might be finally getting where the questions were coming from. Technology is just that, a thing to get done.
Getting a team of people acquainted, comfortable, and productive with said technology is the more interesting part. The heavier responsibility. And the real win.