[cjaybo]

As an audio application developer who still supports users on OSX 10.6.8, I have to ask, what is this about "Windows was the last OS to move to the evergreen model"? Do the breaking changes in MacOS version updates somehow not count anymore?

If developers don't want to support multiple versions of an OS, there are plenty of domains where that isn't an issue. The desktop seems like a weird place to complain about this issue, though, since this is a challenge inherent in the fact that users have choices and freedoms.

Completely disagree with your attitude here.

[JoshTriplett]

I'm not suggesting that apps should drop such support instantaneously or gratuitously. Rather, I'm just suggesting that in the normal course of development, as an OS version becomes sufficiently old and has genuine issues that make support non-trivial, and if the upgrade to a newer version is free and automatic (so it's reasonable to expect people to upgrade), an app developer may at some point say "we expect at least this OS version; if you're using an older version, you're welcome to try, but we don't test on those OS versions so we can't offer any support or respond to bug reports from those OS versions".

I absolutely believe that the "you're welcome to try" part of that is important, assuming there's no known issue (which there may sometimes be). Developers also have an upper bound on available support bandwidth. I don't think apps (or websites) should prevent users from even trying, unless there's some specific technical reason (e.g. a known incompatibility that's producing substantial support burden just to triage, or a library or API that simply doesn't exist on the older version). I do think it's reasonable to say "please upgrade and try again, and if you're still experiencing the issue we'll take a look".

Along the same lines, if a user reports an issue to a website where it doesn't function properly in Chrome 12, or Firefox 9, it's entirely reasonable for the site to respond with "please upgrade, we don't support outdated browsers". It's a little more questionable for a site to say that about a version released the previous month, unless the site is a tech demo for bleeding-edge technology. But at no point do I think a site should actually block users attempting to use older browsers; at most, it's reasonable to show a "not supported or tested, might not work" message.