Re: CLISP EMACS SWANK remote connection crash

C Habs <[email protected]> Tue, 1 Oct 2019 08:06:09 +0000
Newsgroups gmane.lisp.clisp.general
Message-ID <CANe8y8bMhp0-kMJewfRSzHNL3DG3H+xZkmVSGb1v4nVUpCMDsQ@mail.gmail.com>
On Thu, 5 Sep 2019 at 23:41, Vladimir Sedach <[email protected]> wrote:

> > What I do see, as mentioned previously, is that if I remove , (setq
> > slime-contribs '(slime-repl)) or (add-to-list 'slime-contribs
> 'slime-repl)
> > , from, .emacs , then the error handling disconnections do not happen.
> > Putting either of those slime-contribs back in (which seems to deliver
> more
> > features to the  ring-buffer functions) does cause the disconnection when
> > error handling [for an incorrect form].  That is not  major thing to not
> > have - the ring buffer works for the session but history is lost between
> > sessions without those forms in, .emacs - and perhaps the 'solution' for
> > now is to do without.
>
> slime-repl was the whole point of SLIME, and was only moved into the
> optional contrib library in 2008 (I think the rationale was making
> SLIME work for different Lisp dialects). I would not even consider
> using SLIME without slime-repl (my configuration uses slime-fancy,
> which includes slime-repl).
>
> If you change your mind and want to debug the issue, the slime-devel
> mailing list should have some people with advice on how to debug
> SWANK:
> https://mailman.common-lisp.net/listinfo/slime-devel
>
> Vladimir
>

Hello clisp-list

Since the last time, I have attempted, rather weakly I guess, to 'talk' to
slime folk about this.  Early replies from one person suggested to look at
the clisp 32-bit socket implementation/handling and its connection to swank
for the source of the 'unexpected disconnection' error.

In summary, no matter what form is entered that produces an error e.g as
simple as: (+5 6) would do it, the connection is dropped with the
'unexpected disconnection' error message. This only happens when the
slime-repl is is use: in .emacs (setq slime-contribs '(slime-repl)).
Without it, all is ok, but the experience is not nearly as helpful of
course.  Something issue from the slime-repl, clisp, socket connections
then perhaps !

Pertinently though, since I first posted, further testing has cooked this
down to only happening on 32bit slackware clisp, emacs, slime setup. On
64bit, I cannot produce this situation !  I am not talking remote
connections now, just local as such.

Both 32bit and 64bit are the same configs, all the same versions of
everything, same .emacs .clisprc.lisp .emacs.d etc. I've uninstalled emacs,
clisp and slime, rebuilt, reinstalled (as distributed), clean setups,
followed all the instructions on the internet, gone around the mulberry
bush and so forth...but just the same outcome on the 32bit setup.

Would it be a socket problem on clisp, rather than the swank and slime-repl
code, as alluded to,  that is the cause and how can I start to debug this
to see if I can post anything that might lead to pointers of what is not
working, please ?

Regards
Habs

_______________________________________________
clisp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/clisp-list