It’s not for me — I’m on a recent computer that runs Discourse fine. It’s for other users of a forum I’m a part of. Trying to convince them to adopt Linux is probably an uphill battle, lol. It’s hard enough to convince my parents and other people I know to switch (and then you end up having to play tech support for them). I’m not going to try to talk a stranger over the internet into doing that 
But yes, Linux would be a great option for the tech-savvier users of old computers. Sadly, many of those users are on older machines and OSes because they’re not very tech-savvy, or are just tied to their preferences.
They are quaint and old-school that way, but that doesn’t mean I want to exclude them from the communities I’m a part of.
Is this possible even for hosted Discourse Cloud instances? If so, that’d be a great workaround! Seems unlikely though, since that means the Cloud version will have to support multiple versions in parallel…?
First, thank you for the response. I appreciate you (and the team as a whole) even being willing to discuss this.
While you’re absolutely right that older OSes are more vulnerable, sometimes circumstances for individual users make upgrades or OS switches impractical, especially on just a few weeks’ notice (actually a little more than a month, not just two weeks, sorry I miscounted). And it’s not like come May 2nd, users who don’t upgrade to the newest OS will suddenly have their machines join a botnet and explode. Their browsers will keep working, most other websites will keep working, they just won’t be able to post in Discourse forums anymore. Yes, they may eventually succumb to some vulnerability or another, but that probably would’ve been years later than May 1, 2025.
In this case, you’re not really deprecating any particular OS or on any sort of specific timeline anyway. You’re adding three very specific browser features that are not critical to any existing functionality and will have no immediate user-facing impact. Colors can be calculated differently, layouts can be approximated with other CSS tools, and I have no idea what the lookbehind is going to be used for, but there are likely workarounds for that too.
I think comparing these changes to critical OS security upgrades isn’t very fair; they are two completely different classes of changes & deprecations.
But my underlying argument here isn’t about how many CVEs or lines of code any particular change might require. It’s that Discourse is fundamentally about community, and in this case, this decision hurts community for the sake of what seems like a minor improvement in developer experience — but please correct me if I’m wrong.
The cost to the Discourse team to implement a backward-compatible posting mode is measured in dollars and hours. It sucks having to support old browsers, I know, and I and every other web dev all hate it, and I’m sure you do too. However, the cost to your users in this case is measured not just in minor inconveniences, but the very real threat of loneliness and being cut off from communities they’ve long been a part of.
Discourse isn’t just for techies who argue about the relative merits of Windows 7 vs 10 vs the Linux distro du jour; it’s used by people of all ages, across the world, with devices new and old, with different levels of tech savvy. And some of them just don’t keep up with computers and OSes the way we do. Maybe that’s less than ideal, sure, but I don’t know their whole story…
At the end of the day, yes, it’s absolutely true that you cannot support everyone forever, and eventually some software issue will absolutely force your hand and require an upgrade that will leave some small % of users behind.
But those three particular features don’t seem like such an instance. Are they truly so urgent and critical as to justify exiling some of the users who’ve been using your software the longest, with a “too bad, you can no longer post unless you want to do it all by email”?
Those three features really do seem “bleeding-edge” to me, especially for a forum software that’s been doing fine without them for years. Does Discourse really need to be more ahead of the curve than most government, bank, etc. websites? As far as I can tell, they are small incremental conveniences, not revolutionary paradigm changes in software development and maintenance… but please do correct me if I’m wrong and there is some pressing need to implement these ASAP.
But if there isn’t a pressing need… is it truly worth shutting these users out? Those three features seem to have between 91-95% browser support globally (1, 2, 3). Let’s call it 93% on average. If Discourse has 14 million users and this change negatively affects 7% of them… that’s still 980,000 people. Of course that math is simplistic and actual analytics would be better, but the point is that at Discourse’s scale, your changes are affecting real people who depend on these communities for social interaction. Are they truly urgent and necessary? They can’t wait until a basic HTML posting mode is available?