Daemons Now Automatically Reload Config
After making configuration changes, you've historically needed to restart the daemons before the changes take effect. This has always created a lot of support problems, and our various prior efforts to remind or warn users about it haven't moved the needle much.
Earlier this year, we made unrelated changes to how daemon processes work in order to improve daemon scalability in the Phacility cluster (T7352), where we need to run a large number of instances per host. Broadly, these changes consolidated daemons to have a single overseer process for an entire group of daemons instead of an overseer for each individual daemon. While not initially a design goal, these changes made it more practical to automatically restart daemons in a scalable way.
Now, after changes in T7053, we'll try to automatically restart the daemons after configuration changes instead of asking the user to do it. This will probably require some additional refinement, but hopefully represents a more effective approach to this class of problem. Thanks to @joshuaspence for driving the bulk of this change.
I've been continuing work on evolving our infrastructure for editing objects ("ApplicationEditor"), and Paste is now driven by the new infrastructure. There's nothing very exciting or user-facing yet.
The new paste.edit Conduit API endpoint is a preview of one of the things this will enable: the endpoint is driven by the same code that drives the web UI, and has all of the capabilities that the web UI has. Once this deploys more broadly, other applications will generally have similar endpoints which are as capable as the web UI and automatically reflect changes made to the web UI, instead of the endpoints that exist today which are crippled in awkward ways.
I hope to bring ApplicationEditor (T9132) to a basic level of functionality and begin implementing it in other applications in the next week.