[solatic]

It's exactly why taking a trip through the ops/infra side is so important for people - you learn why LTS-style engineering is so important. You learn to pick technologies that are stable, reliable, well-supported by a large-enough people who are conservative in their approach, for anything foundational, because the alternative is migration pain again and again.

[selfmodruntime]

I also feel like we as an industry should steer towards a state of "doneness" for OSS solutions. As long as it works, it's fine to keep using technologies that are only sparsely maintained.

[cyberpunk]

Ingress-Nginx is commonly internet facing though; I think everyone wants at least base image and ssl upgrades on that component…

[rcxdude]

In which case it's even more important that the updates are not a huge amount of work.

[toredash]

I often find myself trying to tell people that KISS is a good thing. If something is somewhat complex it will be really complex after a few years and a few rotations of personnel.

[friendzis]

Quite often the tradeoff is not between complexity (to cover a bunch of different cases) and simplicity (do one thing simply), but rather where that complexity lies. Do you have dependency fanout? It probably makes sense to shove all that complexity into the central component and manage it centrally. Otherwise it probably makes sense to make all the components a bit more complex than they could be, but still manageable.

[hoherd]

Another great one is PLOS, the Principal of Least Astonishment. Stable and reliable software and systems should avoid astonishing surprises.

https://en.wikipedia.org/wiki/Principle_of_least_astonishmen...

[oblio]

At least in the golden days of job hopping, not migrating was a way to hobble that job hopping and decrease your income growth prospects. Now that engineers are staying put more it's likely we'll start seeing what you're saying.

Though now AI slop is upon us so we'll probably be even worse off for a while.