Re: Failing to GC killed buffers considered harmful

[Pip Cet <pipcet@gmail.com>]

On Mon, Mar 30, 2020 at 6:32 PM Stefan Monnier <monnier@iro.umontreal.ca> wrote:
> > FWIW, maybe we should get rid of all_buffers.
> > AFAICT the only reason we need it is to unmark the buffers during the
> > sweep phase, and we could allocate buffers like any other
> > pseudovector instead.

I agree that removing all_buffers would be good.

> The patch below seems to work.

But it changes gcstat definitions (killed buffers are no longer
counted towards total_buffers).

> The only worrisome part I think is that `live_buffer_holding` allowed
> pointers into buffer objects, whereas `live_vector_p` only treats
> pointers to the beginning of the object as a valid reference.

Uh, I don't see how the second part of that statement is true. Can you explain?

> Not sure why buffers would be more likely to have valid pointers into
> them (and can't remember discussions about that either),

buffer text.

> so I assume the
> difference was not important (it just happened to be easier to support
> that for buffers).

Which difference, I still don't understand.

Next, it might be nice to have anonymous buffers that are
garbage-collected automatically and don't need to be killed. *ducks*