Re: [Bug: 21.5-b21] dump-emacs dies for mature xemacs image
Stephen J. Turnbull <stephen(a)xemacs.org> wrote:
>>>>> "Clemens" == Clemens Fruhwirth
<clemens(a)endorphin.org> writes:
Clemens> xemacs-21.5-b21 segfaults, when it tries to dump from a
Clemens> complete xemacs dump. I was trying to cut down the
Clemens> load-time for mh-e by
If I understand what you're doing, don't do that. Although you should
be able to dump from a preloaded XEmacs, you cannot add files to the
existing dump.
I don't want to add files to the dump. I want to preload all lisp
components I regularly use. So, I want to add code to an existing lisp
image.
This kind of preloading is common for all not-just-for-fun lisp
implementations, for instance all major common lisp implementations have
a facility to dump the image. They have the same problem, loading all
libraries/components simply takes too long, even when it's
native/bytecode already.
When you look at
http://www.emacswiki.org/cgi-bin/wiki/DumpingEmacs
it seems that the FSF emacs world also follows this path.
You must start from scratch. You need to add the
necessary files in the right order to the list being dumped. See
lisp/site-load.el for what was the correct procedure five or six years
ago.
I already spotted that, but I can't use the require directive
here. Probably dumped-lisp.el works, but actually I think redumping is
much nicer.
Still, not having any pdump_root_lisp_objects is clearly a bug.
Naive approach: Is it feasable to use the root list as it was at the
time the original dump was made, or do we have to reinitialize it?
---
Fruhwirth Clemens - http://clemens.endorphin.org
for robots: sp4mtrap(a)endorphin.org