[openstreetmap/openstreetmap-website] Use CSS filter instead of dimming tiles (PR #5325)

[Tim Weber]

Let me start by saying that yes, I'm disappointed, but I don't mean to hate. I understand that you're all busy, and that this is volunteer work. I'm not expecting anything from you. Please read my comments more as feedback on how I feel and how things are looking like from my end, written in the hope that the experience might improve for others in the future.

> That's not why it was merged, and I think if you read #5328 you'll see that I'm not "assuming that it's going to fix everyone's problems". That's a pretty unfair characterisation.

AFAICT, #5328 is not talking about #5330 at all. I don't understand how "let's keep dimming the tiles, but just a bit less" is treated as a valid solution to the backlash about the "dark mode" that rolled out yesterday.

And how was that selected as the way to go in the first place? Because #2332 sure doesn't seem to favor it. Instead, there was PR #4712 that brought us here in the first place, merged with no discussion whatsoever, and without pinging any of the existing issues. This feels like "yeah whatever, keep talking, we're gonna do this my way now". And that's massively off-putting.

> It's been less than 22 hours since you opened it. I haven't reviewed this PR specifically, but you can see from the discussion in #5328 that there's lots to consider about the invert+rotate approach. It's not a PR which is a straightforward merge, and even if it was, 22 hours is asking a lot in terms of turnaround from a handful of volunteers.

If this were a feature request, or even a normal bugfix, sure, I understand your argument. But from my perspective, the deployment yesterday _broke the map_ for basically every dark mode user, with basically no option to get back the previous behavior. The only way people could continue using the OSM.org map was either by disabling dark mode for their whole browser (because that's a global setting both in Firefox and Chrome), or use a different browser just for looking at OSM.

That's something that, in my book, should either be hotfixed or rolled back ASAP. And it's also something that wouldn't have happened in the first place if the whole "no we need to weigh the options against each other and thoroughly test and discuss this" attitude would've been applied _before_ rolling out the change, and it feels unfair to _now_ insist on it when the broken release was already deployed.

> I think "failed" is also unfair. There's more to dark mode than just effect on the maps, it took months of work and dozens of changes across that rest of the site too.

And I congratulate you on this. Because I know that it's not easy, and it looked really good. I even acknowledged it at the top of this PR ("The current implementation of dark mode is already pretty good").

But you'll also have to admit that breaking the map for a lot of users is a huge deal. I don't see how the dimmed tiles would've been approved by anyone who cares about accessibility, or uses dark mode. Therefore I have to conclude that it's not been tested or reviewed by the actual stakeholders before going live, and it made things significantly worse for them. I consider that a failure.

That's okay, because failures happen, and then we either fix things or roll the changes back, but instead you're twiddling on some percentages and assume that it's gonna fix the issue, and that's just disheartening.

> #4769 was opened in May, but unfortunately there was no substantial discussion.

The lack of discussion might be related to #2332, the issue titled "Dark Mode", opened _before the pandemic_, the one where people had been suggesting things for years, not being made aware of it.

That the issue has been closed today, with new ones being created for all the different aspects, might make it easier from a project management perspective to keep track on the different topics. But please understand that for the people who have been waiting for five years to see progress on the issue, and now being presented with a half-baked non-solution without their involvement, it probably feels like a joke.

> > not bother trying to upstream any improvements, or contribute to the website going forward.
> 
> That's a shame, we do welcome everyone to get involved.

And I really appreciate you taking the time to address the points I've brought up. I hope that this reply is helpful to you, too.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/openstreetmap-website/pull/5325#issuecomment-2479858361
You are receiving this because you are subscribed to this thread.

Message ID: <openstreetmap/openstreetmap-website/pull/5325/c2479858361 at github.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/rails-dev/attachments/20241115/9606ed8b/attachment.htm>