3 ms·
If version numbers were still sane, this would probably be closer to 10.0.
by userbinator 5y ago
If version numbers were still sane, this would probably be closer to 10.0.
- spicybright 5y agoThey are completely sane. It's just not what you're used to.
- kibwen 5y agoVersion numbers haven't really ever been meaningful. Calendrical versioning is pretty reasonable, and Semver at least makes an effort to give meaning to the version number. But for nearly all software throughout nearly all of computer history, version numbers have been utterly arbitrary. By that metric, I see no reason why 100.0 is any more ridiculous than 10.0.
- unethical_ban 5y agoI think you're both wrong! Parent is wrong that Chrome would benefit from semver, and I believe you are wrong that semver is meaningless. Semver is useful for the kinds of tooling that need minimal marketing, have other software designed against it, and have multiple levels of improvement going on. Some software which are developed against benefit greatly from a x.y.z to let other developers know what class of change has been made. Web browsers and operating systems do not benefit in the same way, because they are a different class of software.
- kibwen 5y agoYou may have misunderstood my comment, as I said that Semver does attempt to make the number meaningful. Though I am not asserting that, on its own, it is very meaningful; it relies on users to define what a "breaking change" actually means, and it leaves it up to third-party tools to enforce any semblance of it. Meanwhile, Semver fails to say anything at all about useful concepts like version ranges. Ultimately, I like Semver! But it's the bare minimum of what it means to have meaningful version numbers, and anyone looking to make them truly meaningful has a lot of work to do. Still better that what we had before, of course.
- superkuh 5y agoVersion numbers of commercial software, or software competing intentionally with commercial software grow rapidly until they become cumbersome for the users to remember and interact with. 100+ is definitely getting there. I imagine a rebranding will come soon with some "Chrome Plus version 1" and a restart. FF will, of course, follow.
- resoluteteeth 5y agoChrome version numbers aren't something that normal users have to know or care about.
- colejohnson66 5y agoIt's not like you can use SemVer on a browser. Too many changes would make it hard to decide whether things are breaking or not. And due to backwards compatibility, it's possible that we'd actually be on 3.100.0 or whatever. It's a lot easier to just bump the major every 6 weeks with releases. Is it the best idea? Probably not, but it's something simple that works and is consistent.
- Pxtl 5y agoChrome does have an API - Selenium uses it. And selenium's chrome driver has strict rules about which Chrome versions it supports as it is versioned. So actually Chrome's rapid versioning does reflect the pace of changes to their API.
- colejohnson66 5y agoBut is Selenium’s requirements due to breaking changes actually existing? Or just a lack of effort to test multiple versions?
- kimixa 5y agoArguably with SemVer as html 1.0 pages still work with no compatibility issues, it will still be 1.x if that is the gate for "breaking" changes. The other way, if you rely on experimental features, it's likely you'll be breaking something every release, ending up with effectively a different major version every release - which is arguably pretty much what we have now. The issue is that for a sufficiently complex environment, the definition of a "Breaking" api change depends on what you use - it's pretty hard to end up making any changes that don't have some visibility across the API after all, even internal details end up leaking.