Re: Preview: portable dumper

[Eli Zaretskii]

> From: John Wiegley <address@hidden>
> Cc: Daniel Colascione <address@hidden>,  address@hidden
> Date: Mon, 28 Nov 2016 12:20:41 -0800
> 
> >>>>> "EZ" == Eli Zaretskii <address@hidden> writes:
> 
> EZ> Thanks for your efforts, but I think this is a wrong direction, that will
> EZ> eventually get us into the same problem we have with unexec: the need to
> EZ> know and depend on intimate details of relocations, memory allocation,
> EZ> etc. On top of that, adding Lisp objects will now require writing its
> EZ> dumper back-end, so this will be a constant maintenance burden of the kind
> EZ> that only a few of us can bear.
> 
> Is there a way to get away from such a requirement, Eli?

Yes, by loading Lisp code directly.

> If I understand Daniel's contribution, he's proposing the equivalent of a
> program loader for Emacs Lisp byte-code, stored within what amounts to a
> custom object file format.

No, it's not only the byte code, its the entire state of Emacs after
loading that byte code and initializing variables and subroutines not
in Lisp.  IOW, all of the Lisp objects in the entire heap, no matter
where they came from.