“or pushing an “update” that’s really a rollback once enough people have gotten the update to reveal a bug that was missed before release?”
This is certainly possible for purely code issues but sometimes environmental issues (caused by the previous update) are harder to fix without over-stepping your bounds. You can’t just go deleting files off the client’s system! More often new features in Firefox have an ‘off switch’ in about:config, so disabling it is as simple as a toggle.
Fx has so… many… freaking… unit tests. I spend more time writing Fx unit tests than I do writing patches. There are pros and cons but at end of the day it’s about making the best end product, so total code coverage is a beautiful thing.
The automatic updating is not available on Linux editions of Chrome - it is the responsibility of the user to keep the program up to date using their package manager (such as APT or Yum). The browser itself has no means of updating itself (since this would require root privileges).
Secondly, it actually is theoretically possible for a program to update itself while it is running - although a currently-running executable cannot be deleted / modified, it is possible to modify the binary machine code in place. Obviously this would be horribly complicated and introduces all kinds of potential errors - but I just thought I’d mention it.
Interesting to note that the spammy SEO attempts of the rc helicopter commenter above have failed to get the site in question anywhere in the first several pages of a Google search for “rc helicopter” as far as I can see.
This is great for consumer uses, but big corps absolutely hate not having every version of software under their control. One can argue about whether it’s right for them to do it, but I have worked in too many corps where upgrading to IE 7 broke the 50 vendor-provided apps that depended on IE 6. Vendors aren’t in a hurry to upgrade, and if “it just works” then the corporations won’t either. Even Firefox has trouble in corporate because of their upgrade policy. Until the old-school corporate IT mentality of “well the app is certified on X browser” goes away, don’t expect to see chrome, with the changing version all the time, every make much inroads into corporations.
My issue with the Infinite Version is when you start depending on systems that work. While you may roll things back quickly, you still have unavoidable downtime if you roll out an update that breaks the browser and the autoupdater.
What I don’t get is this: Do they calculate diffs only from the previous to the current version? What happens if I missed a couple of versions? Will I receive the diff from my version to the current or will they push me each diff separately?
Hard to believe we’re nearly at Chrome version 100 these days… and as I said in an earlier blog post:
Chrome is a joy to use, and in my opinion at least, it’s the first true advance in web browser technology since the heady days of Internet Explorer 4.0. Chrome is filled with so many thoughtful details, so many reimaginings of web browser functionality as a true application platform, it’s hard to even list them all.
They probably should have used a letter after the first digit. Then they could have had 27 versions before moving on to #2. (#1, #1A, #1B, etc.). But who knew?