[Sbcl-announce] sbcl-1.3.0 released

[Christophe R. <cs...@ca...>]

Dear all,

New evaluator (even if it's disabled by default), new platform (even if
it's a bit difficult to get hardware for... it seemed like a good
opportunity to have a new minor version number.  I've just released
sbcl-1.3.0, and uploaded a source tarball to Sourceforge.  I'm including
a summary of changes at the end of this message; please consume
responsibly.

Best,

Christophe

changes in sbcl-1.3.0 relative to sbcl-1.2.16:
  * minor incompatible change: the environment passed to a macro/setf/deftype
    expander is not always an object of type SB-KERNEL:LEXENV.
    It can be nil, as is permissible by the standard.
  * enhancement: ported to ARM64 Linux.
  * enhancement: a new interpreter is included which has many benefits
    over sb-eval. It is disabled by default. See src/interpreter/README
    for instructions to enable it, and further details.
  * bug fix: calling COMPILE when SB-EXT:*EVALUATOR-MODE* was :INTERPRET
    would fail to perform "normal semantic processing such as macro expansion"
    as stipulated by X3J13 issue LOAD-TIME-EVAL.
  * bug fix: SB-UNICODE:CONFUSABLE-P no longer treats "<" and the empty
    string as confusable.  (lp#1504739)
  * bug fix: (ASSERT (COMPUTE-IT ...)) would expand incorrectly
    if COMPUTE-IT was a local macro that shadowed a global function.
  * bug fix: SB-POSIX:CFSETISPEED and SB-POSIX:CFSETOSPEED now work properly
    on platforms that use only the minimum specified set of fields in struct
    termios (verified on Linux, still known not to work on FreeBSD and NetBSD,
    very probably others).  (partial fix for lp#1500951)