[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: In 17.7, . . .
- To: (BUG LISPM) at MIT-AI
- Subject: Re: In 17.7, . . .
- From: AGRE@MIT-AI
- Date: Sun ,20 Aug 79 05:26:16 EDT
QC-FILE seems to have a very random bug which, as close as I can tell, goes
like so: when the input file to QC-FILE has a readmacro character in it,
e.g., as in (MATCH '(?X Y Z) ARG) where "?" has been defined by
(SET-SYNTAX-MACRO-CHAR #/? #'(LAMBDA (X Y) (LIST '/? (READ X Y)))),
the file reader sometimes causes the READ to point at the Lisp Listener
rather than at the file, causing the compiler to hang until the read
is satisfied. The evidence I have to support this explanation of the
problem is that when I do a QC-FILE on my file AGRE;BWASST >, which
contains "?" characters which are supposed to be expanded as mentioned,
the compiler hangs and "TYI" comes up on the job line. If I answer
with (FOO BAR) <carriage-return>, it goes away. However, the compiler
then complains that (FOO BAR) is not an atom (the actual form which was
mis-expanded was ?PROP, which eventually causes PROP to be SETQ'ed), and
exploration with the debugger reveals that the macro expanded as though
not ?PROP but ?(FOO BAR) PROP was in the source code.
I would greatly appreciate it if this problem were to be fixed, since
17.7 is loaded on cadr-2, and I have just finished overhauling my code to
make it compatible with 17.7. If I have not given you enough specifics,
I would be glad to arrange a demonstration.