David-
Post by David RushIt seems that if I examine a cyclically linked data structure in
Larceny's debugger, I will never see another command prompt. This
turns out to be mind-bogglingly annoying, and I was wondering if it
was a known bug, or if there was some setting which might ameliorate
it?
I recently had to wrestle with this behavior when interacting at the
REPL.
I suspect it qualifies as a known missing-feature, rather than an
outright bug [1]
----
There is a print-circular library in lib/Experimental. In my case, I
was able to load that up and then hook it into the repl-printer
parameter, but that does not solve your problem, because the
debugger's printer (which is distinct from the REPL's printer) cannot
currently be overridden by the user.
Perhaps we should make the debugger's printer a user-settable
parameter. Do you think that would suffice for your needs?
(The debugger infrastructure really is pretty minimal, for better or
for worse; you might be just as well off in the short term making your
own copy of the source code in lib/Debugger/ that uses print-circular.)
-Felix
[1] In my ideal hypothetical world, cyclic structure would be
transparently handled by the write procedure that the debugger is
already invoking. My impression is that the current (but undesirable)
behavior (of Larceny's write procedure) is allowed by the R6RS, but am
not sure (questions arise about whether a cyclic structure is a datum
value as described in section 1.10; assuming a cyclic structure is not
a datum value, then the next question is how to interpret the word
"should" in the first sentence "Datum should be a datum value" in the
description of the put-datum procedure, which serves as the
specification of the write procedure in the R6RS.)