On Tue, Mar 18 2025, Rutherther wrote: >> As fuel for thought, we have https://issues.guix.gnu.org/27155 from >> 2017 which creates a finalization mechanism for any service type, but >> it built on the idea of explicit extension points which would limit >> what a finalizer can do to whatever a service writer has made >> available. > > Could you clarify this point?
I have either misinterpreted, or miscommunicated. You're right that the initial patches don't rely on explicit extension points. The discussion moved on to talk about explicit extension points, though, which is what I was commenting on. It seems like there wasn't consensus on a global finalisation service either way. I'm now thinking that this should be a field on operating-system instead of a part of our service mechanism. That way we don't have to worry about services arbitrarily transforming each other, but we still allow the operating-system author to modify their services as they wish. Carlo
