abort: instantiation fault in writeln_body(X, T2, eclipse)

From: Christian Fritz <Christian.Fritz_at_post.rwth-aachen.de>
Date: Sat 27 Mar 2004 01:33:05 PM GMT
Message-id: <40658291.1050601@post.rwth-aachen.de>
Dear Eclipse-Users,

We keep getting above error and cannot make out why.
The entire program is much to complex to present it here
as a whole, but what we are principally doing can be
sketched as follows:

cfunction1(X1), writeln(X1), cfunction2(X2), writeln(X2),..
cfunctionN(XN), writeln(XN).

where cfunctionI(XI) is a call to an external C-function binding
variable XI. We call this goal repeatedly sleeping about 100ms
between two calls and most of the time it works just fine. However, every
once in a while eclipse aborts with an error like
"instantiation fault in writeln_body(XI, T2, eclipse)" where the
"I" varies. We already made sure, the C-functions always succeed
and do not unify any NaNs or INFs (which some days in the past
caused us some segmentation violations). Now the most confusing
thing is: the output of writeln(XI) is printed to the screen prior to 
aborting.
That is, it seems as if the instatiation fault is raised during some form
of backtracking. This in turn is weird, as the only way of producing this
error is doing something like: "writeln(Stream, X). " that is, naming an
unbond variable as the stream to print to (note that we are using
writeln/1 and not writeln/2 as the error suggests).

Maybe of relevance for this problem is, that, independent of above goal we
use a timer of 100ms (set_timer(real, 0.1)) to call another C-function
(within an appropriate interrupt handler).

After three days of intensive debugging efforts with absolutely no
success, we hope you can help us or give us some other hint where
to look at.

Christian
Received on Tue Mar 30 04:48:30 2004

This archive was generated by hypermail 2.1.8 : Wed 16 Nov 2005 06:07:28 PM GMT GMT