Re: Reachable killed buffers

[Stefan Monnier]

> Since our discussion shows that there is no solid and bullet-proof
> solution for killed buffers problem (IMHO), I would like to propose
> an alternate (less effective, but safe) solution:

I don't think this issue deserves that much effort.
I think the original behavior (of keeping some buffers longer than
needed) wasn't that terrible and the current behavior (of sometimes
removing deleted buffers but not always) is fine as well, I suspect that
in practice those lists are pretty much never reachable from elsewhere,
so the "risk" of not removing all deleted buffers is very low.

I.e. let's keep the current code and move on to more useful issues,


        Stefan