Vadim,
Very good! Thank for your excellent detective work on this problem. I
have no problem with using the latter form below for invoking maxima
from xmaxima. The server option was designed to make things simpler. If
it doesn't make things simpler, by all means feel free to bypass it.
--Jim
On Tue, 2004-03-02 at 00:06, Vadim V. Zhytnikov wrote:
> I've been trying to understand why new xmaxima doesn't work to me
> with all lisp on Linux and it seems that I've got a rough idea
> about what is going on.
> There is only one subtle difference between new
> maxima -s PORT
> and old
> maxima -p server.lisp -r ":lisp (user::setup PORT)"
> way to start maxima server. In the new scheme socket
> connection is initialized at the very beginning of maxima session
> before maxima banner is printed. While with old scheme
> socket is initialized _after_ maxima banner.
> It seems that xmaxima read-eval-print loop is too fragile and this
> banner confuses xmaxima. In particular if I replace
> standard maxima banner by some "bla bla bla" then new
> xmaxima works with gcl without trouble. Unfortunately this is
> not the case with cmucl. xmaxima -l cmucl hangs even if
> I make maxima banner empty. Certainly the issue can be
> debugged further and the right resolution is to make xmaxima
> read-eval-print loop more robust with the help of
> *prompt-prefix*/*prompt-suffix* etc. But I venture to
> suggest a workaround - partial fallback to older scheme
> in xmaxima
>
> < lappend command -s PORT
> ---
> > lappend command -r ":lisp (setup-server PORT)"
>
> From xmaxima point of view we return to old scheme
> and it works to me on Linux with all three lisps -
> gcl, clisp and cmucl. Maybe it will help to resolve
> xmaxima problem on Win 95 as well. I strongly believe so
> since old xmaxima used to wok well on Win 95 (well Win 98SE
> and Win Me I newer tried maxima 5.9.0 on Win 95 but I'll be
> surprised to find any difference).
>