Have you ever read aloud Python 2.5 as "two and a half"? I did. My grandma would read it like this as well. Why would you expect a typical end user (in 2+ billion world) to read is as "two dot five"?

We can clearly see major end-user applications steering away from semver. Chrome and Firefox, for instance, currently use single-number versioning and already reached ver 100+. What does "breaking change" even mean for end-user applications? Some developers don't like huge numbers after a dot (looking at you, Linux), and increment the major version arbitrarily.

15 years ago I have seen some instances of decver and laughed at it, but now I think: why not...

Consider an online game updated quarterly. The previous releases were 6.7, 6.8, 6.9. What version number comes next?

6.10... Of course 7.0 it is! Cannot you spot a simple arithmetic sequence? Even little children know that 6.10 < 6.9. (And some wonder why the trailing zero).

The main assumption of decver is that when normal people see numbers, they assume decimal. The core rule states that:

verA is "newer" than verB <=> float(verA_string) > float(verB_string)

The rest is to be decided individually in each case.

1. Plan major milestone releases as most yearly or bi-yearly. That will increase the major version (number before the dot).
2. Release at most 9 main preview versions, where the decimal part represents feature progress towards the milestone.
3. Any minor fix releases only slightly take the system towards the goal, therefore only slightly increase the number. E.g. 4.7 -> 4.71.
4. Version release candidates with a sequence approaching the next integer. E.g. 4.9 -> 4.95 (aka alpha) -> 4.99 (aka beta) -> 4.999 (aka rc-1). Finally release version 4.(9) == 5.0.

TONDO

• TeX
• Tibia

Kinda...

Just use YYYY.MM versioning system...