[Openmcl-devel] *read-suppress* REPL reader "bug"

[Daniel Herring]

Bug in quotes since it appears to be entirely conforming and common, and it
falls into the "rm -rf /" or "deltree c:" class of stupid.

If at the REPL you type (setf *read-suppress* t),
CLISP: ignored
SBCL: useless REPL
ECL: useless REPL
CCL: useless REPL
...

I couldn't get "useless REPLs" to return anything other than nil.

While dpANS/CLHS don't seem to say anything about this, I believe CLISP has the
right idea.  The REPL should (lexically) ensure *read-suppress* is bound to nil
before reading user input.

- Daniel

P.S. I separately sent this message to sbcl-devel, ecls-list, and
openmcl-devel to avoid automatic cross-posting of replies.  Other
implementations may also be affected.