slime + clisp + emacs: server disconnects

[Vladimir Sedach]

> Perhaps I can hack the swank loading procedure to avoid this, for
> instance.But I'd need to understand the situation beyond the simple
> observation that there's a crash, hence the mailing list message.

If CLISP crashes because of a bad memory reference (SIGSEGV), then it
is a problem with the internals of CLISP, which is triggered by SWANK
only incidentally. Even if there is some workaround for loading
SLIME, it will not be very useful, because CLISP will continue
crashing on whatever code you want to work on.

You need to trace down the point where it crashes by stepping through
the SWANK code in the debugger (you can narrow it down quickly by
putting break or print statements in the code in a binary search
pattern), isolate the problem code in a self-contained, reproducible
example, then try to debug the CLISP run-time, by re-compiling CLISP
with run-time debugging support and using GDB/strace/the usual C
debugging tools:

https://clisp.sourceforge.io/impnotes/faq.html#faq-debug

There is a CLISP mailing list where you can get more specific
information:

https://sourceforge.net/projects/clisp/lists/clisp-list

Vladimir