Nice article, thank you! I feel all the pain =)

For large applications with a lot of technical debt, this is a lost race if you can't convince the business to hold a "core/infra" team that will be responsible for keeping everything up to date... and even so, with 20-30 deps, which is not much for a large application, this is an everyday task.

Plus, you might get a surprise out of the blue. Well, not suddenly, but one day, in a very large, old and complex application, I got a build surprise. The Angular team broke passing some parameters from the underlying webpack to the even deeper tsc, and this was important because the build pipeline had around 10,000 LoCs to keep all the legacy >1M LoC together.

Of course, I didn't have any changelogs or announcements about it because it was too deep technical and you should f* off and use the mainstream (I'm kind of proud of how I was able, with such meager resources, to manage updates to such a huge chunk of, well, mess, from Angular 1.2 to Angular 7 in all versions... jQuery, Bootstraps, Charts... highly customised and abandoned components like selects and date pickers... well, I have something to re-watch in my nightmares...).