> But then I'd say it is even worse, as you're informing desktop.el
> about an implementation detail of eglot.el. If I change that minor
> mode's name, then I have to change desktop.el as well.
That's okay: it's desktop.el's job to know about some implementation
details. Just look at how much it knows about what the various modes
and variables do in Emacs.
Wait, you're saying it's "okay" to have to do a commit to Emacs's repo
everytime someone makes a third-party package that has a minor mode
that needs special handling? Or everytime someone changes the name
or shape of a minor mode?
I can't possibly see how this is okay. There should simply be a generic
mechanism for minor-modes to tell desktop.el and other intrusive
packages to "stay out of my minor mode".
But we do have that mechanism. It's called symbol properties and it's a nice
feature of lisp. So let's use it, please. All the other solutions are demonstrably
worse. I'm not even saying to get rid of d-m-m-table, as Lars is. I'm just
saying: let's not type "name-of-eglot-symbol-that-joao-may-want-to-change"
into desktop.el. It's really a hazard.
João