Hacker News Reader

Comment by influx | Hacker News Reader

Firefox made a huge mistake following in Chrome's trail with the version numbering scheme. I just don't care about Firefox 76, is it a big change? Is there some new feature I would want to try? Who knows, I'm sure Firefox 77 will be out soon though!

View on HN

I want to figure out who to talk to on both Firefox and Chrome sides to beg that they adopt a scheme like YYYY.MM.minor for their version numbering. I believe that would be superior for basically everyone—web developers, users, &c. (In the case of Firefox, it works especially well since they’re releasing approximately monthly now. I could imagine people on Chrome puzzling over why some month numbers got skipped—not that people actually look at those numbers often.)

Might affect (dumb) uses that rely on the current increasing version number scheme in user agent strings and package/deployment scripts.

User agent strings are progressively being frozen, so that shouldn’t affect anything.

if you strip out the periods, it’s still an increasing numbering scheme, assuming there’s fewer than 10 minor releases in a month, which seems highly likely.

Try talking to the guys who keep "improving" location bar. Maybe that will keep them busy for a few releases.

I agree. The first thing I think when I see a version is "how old is this?". Version numbers as they are now are completely arbitrary (nobody really follows semantic versioning no matter how hard they try). Might as well encode some useful information into them. I'm thankful a lot of the Linux stuff I use follows this.

Releases are almost always date-based, but once in a while a release is delayed. I guess that's a bit easier if the release number is not a date.

There's also the fact that Firefox is now on a 4 weeks release cycle, which means there will be months with two releases. (and that's ignoring chemspill releases)

Then if not year.month they can use year.counter where counter goes from 1 to 14 = ceil(366/28)

i mean, whats the problem with the version going up 2 numbers rather than 1?

It's suboptimal to have to change the number of the next release during a release cycle. That would mean, for example, that Firefox Nightly 2020.11 might correspond to a Firefox release 2020.12.

Aw yes, the Ubuntu style.

What’s the advantage behind a more complex scheme? It doesn’t seem to add anything. The current system just notes the specific release in a series of releases.

Semantics. The main thing that users ever want on seeing a version number is to know when it corresponds to. That’s a bothersome lookup at present.

You always want to be on the latest one, because that's where the security fixes are.

And if you care about what's in it just read the release notes.

But in real life as a developer, not all of your users are using the latest one. If you depend on a certain feature, it’s a pain after looking up the version that started supporting it, to then have to look up what time that number corresponds to. (e.g. “that’s last month’s browser” is a bit different from “that got fixed in the browser two years ago, please update”).

In real life, developers use https://caniuse.com which tells you exactly what features are supported in what version and when that was released.

yea. you support features / ages / percentages, not version numbers. whether there are 2 or 20 doesn't really matter.

A moment where version numbers are relevant is when you collect telemetry from your users and when you reccomend supported browsers.

I think it is in part a difference between websites oriented to the public and websites oriented to companies, where browser versions can be more complex than a self-updating browser.

Sorta. Though in that case, more versions is more better - it gives you more precision on the up-to-date-ness of the browsers people are using. Rather than saying "80% of people use a browser from between 2016 and 2018", you can say "75% use one from late 2018".

You don't have too. There is Firefox ESR.

You are aware of the long term support firefox esr, right? Because the existence of that invalidates the "only the latest gets security fixes!!!" argument.

looks like v68.8 is ESR... not far behind, i was hoping for v10

Not if an update renders half of your add-ons unusable. Which happened several times in the history of Firefox (I'm using it since Firefox 2).

I prefer date-based version numbers so that I can easily tell how out of date something is - assuming it's updated frequently enough.

But either way, I don't think this is a particularly important issue.

How is that addressed by a more traditional version number? There are many examples of major software upgrades reflected by a change in minor version number, minor software updates reflected by a change in major version number, or a major new release having less functionality than the previous version (e.g. Final Cut Pro X). With mature software like Firefox, what even is a "major change"?

Calling it Firefox 13.17.6 is not an improvement.

Firefox and Chrome already release on a fixed date schedule instead of a feature based schedule. The proposal is to simply use the date in the version directly instead of indirectly.

The commenter at the top of this thread wanted to know whether this version of Firefox is a "major release". Using the date doesn't change that.

"Firefox 10-04/20” "Firefox 10-05/20” then " Firefox 11-06/20”. Though deciding what's a "major release" seems like a thankless hassle.

Indeed.

There is no right way to decide what is a major release. Whatever you do, some users will think a major release should have been classed as minor, and vice versa. It's ultimately easier for everyone if you don't make the distinction.

"Major release" also doesn't matter like it used to. One of the big new features in recent versions of Firefox is WebRender, but that is being slowly rolled out among users based on CPU/display/platform, it's not a Big Bang feature release.

Firefox 76 or Firefox 5.6.6 makes no difference to me really. With browsers everyone should aim to be on the latest version and definitely not very old versions. To know if Firefox 76 is the latest or really old you need to get some more information. I wonder if anyone has tried to do it as follows:

Let the latest version be called Firefox and then adjust the previous versions to Firefox -1,-2,-3,-4,-5.. -76

You are like _eight_ years too late with that discussion ... trust me, it has been extensively covered :-)

And yet we still have an arbitrary number that is no better than just sticking a release year+month on it, though it is less informative.

I have been saying this from the day this was announced, so not sure this is "eight years late" or an eight year old unresolved bug.

If they follow Ubuntu, I'd wish they name it Cinco De Mayo for this version.

Reply on news.ycombinator.com