* EDIT: In the sense of being non disruptive and safe; the features are very non-boring:)
If there is one thing I've learned from reading various messages from Linus on the mailing lists, is that WE DO NOT BREAK USERSPACE!
So Linux would be perpetually on the same major version, which means there is little reason to use Semver in this case.
Also, the first number gets updated whenever Linus feels like the second number gets too large, not sure how "2.1028.0" would solve that :)
Yes:)
> which means there is little reason to use Semver in this case.
Okay, that's fair. If semver is there to signal breaking changes and there will never be breaking changes maybe it's not super useful.
> Also, the first number gets updated whenever Linus feels like the second number gets too large, not sure how "2.1028.0" would solve that :)
That's not "also"; I'm specifically saying that Torvald's versioning system is bad. Honestly I'd be happier with losing the leading number completely and doing an increment per feature release; right now the "major" version field just doesn't mean anything.
Not that most users care, as most of the tens of thousands of devices supported are fairly obscure and the changes don't affect the other users… yet kernel API changes that break, say, Nvidia's drivers or ZFS or other high-profile out of tree drivers, have an outsized effect. You either have to increase the semver major version with every single release, or somehow version internal APIs separately, which isn't really practical either.
The problem i see is, what's next? Google chrome is at 106... a few more years, and 200.. 300? What happens if linus runs out of fingers for the major number too?
I miss the "good old times", where 2.4.17->2.4.17.1 meant a minor patch, ->2.4.18 meant a new version with some new features, -> 2.6.x meant some large changes, and 3.x meant something huge, with possible backwards incompatibilities and making you choose, between a mature 2.x version vs a newish 3.x version.
One name for the marketing brochures, one for techies (and their scripts). Works for me.
This massively predates web browsers with 100 versions
Also in 2011 firefox went from "slow versioning" (up to version 4) to a new major version number every few months... the unofficial reason was, that chrome was "overtaking" firefox in version numbers, and "higher number == better browser" (chrome went from 9 to 16 that year).
Now, if Google would also enforce open-source compliance... the amount of hoops one has to jump through to get OSS dumps is absurd, and most of these lack critical parts (e.g. build toolchain, kernel config, partition layouts, DTBs) so it's OSS in theory but unusable to develop alternative OSes in practice.