[BUG] Respect SemVer (not a duplicate - please read)
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Claude Code versions do not follow Semantic Versioning. Every release appears to be tagged as 1.0.(x+1), regardless of whether the change is a bugfix, a new feature, or a breaking change. This makes it impossible to know if upgrading is safe, or if it might introduce new functionality or breaking behaviour. This is helpful for security (users can feel free to always update to the latest patch version) as well as user experience (I will read patch notes on minor releases to see what new capabilities are available).
There have already been two issues raised about this, namely #5869 and #7153. However, they were simultaneously closed as duplicates of _each other_, which leaves the concern unresolved and no open issue to track this. So please do not close this as a duplicate, as there is no such open issue at the moment. Other related issues include #5770 and #5735 although these ones are specifically about the CHANGELOG, not SemVer.
(I'm hoping by explicitly tagging this as "not duplicate", the automatic issue sorting tool won't close this... 🤞 cc @mathiaswillburger)
What Should Happen?
Releases should follow SemVer so that:
x.y.(z+1)= bugfix (safe to update)x.(y+1).0= new feature (safe to update, but might introduce new bugs)(x+1).0.0= breaking change.
If SemVer is not respected, there should at least be clear documentation of what versioning scheme is being used. If following this is not desired, I recommend doing (x+1).0.0 at the very least, or x.(y+1).0 if it's possible to commit to non-breaking changes between releases.
11 Comments
I really hope this issue remains open and gets further consideration.
My related issue was just automatically locked after being closed. The GitHub actions might be a bit too aggressive in deduplicating content.
Friendly ping @bogini. The claude code version is currently v2.0.25 (and before v2, it was v1.0.19 or something). I'm pretty confused by this. What does the claude code team plan to use the middle number for?
@bogini anything on this? it's also related to my https://github.com/anthropics/claude-code/issues/12071
@Jarvvski you make a good point in #12071! Rewording it with this context - rather than just standards (as originally motivated this issue), having the minor versions in place has the practical advantage that Claude Code wouldn't need to send so many notifications for patches; only for new features from minor releases.
I admit I do not understand the motivation for
x.0.yversions exclusively. Perhaps it started as a bug and it just stuck. But it would be nice to have proper semver! @bogini (or whoever is in charge of this)This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.
Still an issue
This issue has been automatically closed due to 60 days of inactivity. If you're still experiencing this issue, please open a new issue with updated information.
Still an issue
I won't comment on the practicaly feasibility of sevmer (see e.g., https://gist.github.com/jashkenas/cbd2b088e20279ae2c8e, or a talk by Rich Hickey https://www.youtube.com/watch?v=oyLBGkS5ICk#t=13m47s) but I wanted to let you know that:
Weird. Thanks.
I guess we might need to email someone at Anthropic? The issues seem over capacity so maybe they can't check them.
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.