Semantic versioning. If I have 1.0.0 and you release 1.1.0 I can be pretty confident it's safe to update. If you release 2.0.0 I need to read the release notes and see what broke.
If I have version July2023 and you release August2023 I have no information about if it's safe to update. That's terrible. That's really bad.
This is for dependency management and maybe apis more than OSs, but in general semantic versioning is a very good system. It should be used often.
I'm partial to semver where it makes sense and date based releases where it doesn't. At my work we use <year>.<month>.<version> like 2023.7.v2 for template releases but semver for apps with APIs and such
X is for major overhauls. Y is for a new individual feature added or dramatically reworked, Z is for bug fixes, updates and polish.
Like Blender is currently on 3.6. They had a dramatic major program wide overhaul a few years ago. And since then have been adding new features and reworking old ones in major 3.X releases, and occasionally have smaller updates and fixes in between, giving us 3.X.Y updates.
The only thing I don't like about that versioning system is the ambiguity that can sometimes arise due to different interpretations of what the numbers after the first dot mean.
You could either say:
It's a decimal system, therefore 3.4 is bigger (comes after) 3.13. (3.4 > 3.13) or,
The numbers after each dot are independent, therefore 13 is bigger than 4, so 13 is the newer release.
It's usually fairly obvious from changelings but every now and then I get tripped up.
I thought Linux Mint did this, but apparently they're kinda fuzzy about it? Which was not great to learn when I went to update an old laptop, and briefly thought the project had just died.
I had to type this three times because Lemmy closes the comment box and dumps whatever you had typed, if you upvote another comment while it's open. That's objectively terrible.
I had to type this three times because Lemmy closes the comment box and dumps whatever you had typed, if you upvote another comment while it's open. That's objectively terrible.
Yikes, that is terrible. What client are you using?
I was looking a Linus/Linux comment, I was trying to remember at what point Linus said "I'm incrementing the major version because these numbers are getting too big, there is no major advancement".
NT was a parallel line of "professional" windows. It had a different kernel or something. There were equivalent versions to most of the home releases.
The first release was NT 3.1, to match version numbers with the home OS.
NT 4 was the professional version of win 95/98.
In the year 2000 Microsoft released both Windows ME, and Windows 2000. ME for the home, 2000 was the NT release for the workplace.
The products were merged with windows XP, now all windows is windows NT.
The version numbering makes sense if you count by the NT version numbers. 2000/ME is version 5, therefore XP is 6, and if you pretend Vista never existed (as you should for your own sanity) you get to windows 7 and it all starts to make sense.
Different kernel. 95 was still DOS based. I believe a significant amount of stuff (especially drivers of course) which worked on one side didn't work on the other.
XP was the "merger" - the first NT based system for the consumer market.
Decades long projects during which time the thought on when you should change the version number and what a version number even is has changed multiple times.
I know id got on the anthology-naming thing after Quake II, but... FEAR was right there. Pick another monosyllabic name for your id-formula FPS and half the criticism would vanish.
Basically, they were doing 3.15, 3.16, etc but they decided to turn it into whole numbers. It was currently 3.19 so they decided to go from 3 to 4.0, and remove the decimal.