I recently ready this really good post titled Cold-blooded Software, and it got me thinking about the two primary reasons that I tend to avoid adopting new software:
- There's a semver major change coming soon
- The software has become abandoned
The second feels reasonable, why invest in something that is no longer maintained, no longer gets bug fixes or new features, or nowadays more importantly keeps up to date with the other libraries and frameworks that it gets adopted alongside. A recent example of this was that I wanted to look at adopting Contentlayer for my personal blog, however I checked a few open issues and noticed that the core maintainer of the library hasn’t been active for a number of months at the time.
My immediate reaction was that I didn’t want to adopt it because it might not keep up to date with other features or libraries that I used in my personal blog.
However, this isn't the point of the post, instead the first reason above I haven’t really seen much content be produced about, but it honestly keeps me away from libraries and frameworks far more often than the abandonware case, and it seems like it impacts others just the same:
Spending time and effort to invest into a library or a framework (or a language for that matter), only to have that be turned on it's head following a semver major release is disheartening.
You feel like you're making progress, beginning to understand the system, it's API, the extents of the system, even it's trade offs, and then you need to throw that all out the window and start over.
Even for libraries that pride themselves on making minimal breaking changes get burnt by those that don't. Someone gets burnt by the Common JS to ES Module migration on NPM, and now they feel like the semver major release of Next.js from 13.x to 14 is going to impossible to work through[1].
I'm not sure what the solution is here, maybe myself (and others) can get more comfortable software changing. Or, just maybe, we can start to harden our software, make it more slow moving and slow updating, and also maybe vendoring in more things instead of depending upon as many libraries as we do.
