Notices by Jason Francis (jf@hachyderm.io)

@codewiz I think I am understating just how broken the old APIs were. They are unworkable. Like they are so busted that there is no way to transition them without breaking them further. They just don't work. This was all written up here a few years ago: https://blog.tingping.se/2019/09/07/how-to-design-a-modern-status-icon.html

Sure it would have been great to have the design of the background apps API twenty years ago! But it took lots of years of experience to just get to this point where it could be designed.

@codewiz And BTW this is what I was getting at. One of the reasons why it was important to just remove it early was to *avoid* this situation where some unworkable API gets dumped into someone's lap who didn't ask for it, and they are stuck trying to figure out how to transition away from it forever.

Pipewire is an interesting example because there are actually many broken edge cases with Pulse compat that likely won't ever be implemented, you can see them if you check the Pipewire bug tracker.

@codewiz @dropbear42 @globalc That was never "killed" it was always trivial to add it back with extensions, many distributions did that. However this has arguably caused worse problems because every API for it has been pretty flawed and caused broken behavior in apps. So I don't think it was premature, it likely should have been removed sooner.

The background apps API is the only one that actually works correctly but it took this long just to design it.

@dropbear42 Yes, that is why I refer to distributions, e.g. this was supported by Ubuntu and a few others so they would keep the extension working whenever they did a release with a new GNOME version. So you could be reasonably sure an Ubuntu extension would work elsewhere.

But even so, this was still trivial to keep working, I don't think the shell extension API for indicators has actually changed at all since the first release of GNOME 3 despite being considered "unstable"