2023-09-10  OVERCOMING "LISP CONNECTION CLOSED UNEXPECTEDLY"
                     BY CLEANING .CACHE/COMMON-LISP

Cleaning up whatever is in .cache/common-lisp helped me get around
issues that began with the update of slime and swank to version 2.28 in
Debian unstable.

Using slime would invariable go something like this:

,----
|  PreviConnecting to Swank on port 42941..
|  Connected. Let the hacking commence!
|  Open my-obviously-buggy.lisp
|  Lisp connection closed unexpectedly: connection broken by remote peer
`----

Whenever I wanted to work on an obviously inperfect Lisp project of my
own.

This would also happen when running

,----
|  emacs -q --debug-init
`----

So it seemed like this wasn't something in my local settings.

The *inferior-lisp* buffer would show message like these (redacted and
formatted):

,----
|  Backtrace
|  SWANK DISPATCH-EVENT SWANK MULTI THREADED-CONNECTION
|  FLET SB-UNIX BODY IN SB-THREADRUN
|  FLET WITHOUT-INTERRUPTS-BODY
|  SB-THREADRUN
|  foreign function call_into_lisp
|  foreign function funcall
|  Connection to Emacs lost.
|  condition dcase failed WRITE-DONE 2
|  type SIMPLE-ERROR
|  style SPAWN
`----

Searching in the Slime issues[1] tracker eventually encouraged me to
clean up m local .cache/common-lisp files. And that seems to have
resolved it. There was lots of old stuff in there, going back to SBCL
2.0.1 from around 2020, and SBCL is now at 2.3.7.

Hope this helps others.


[1]: https://github.com/slime/slime/issues/