sbcl-1.5.7 release; more details about SBCL20 in Vienna

Dear all,

The Wheel of Time turns, and Months come and pass, leaving memories that
become older sbcl releases.  Older sbcl releases fade to unsupported
status, and even those are long deprecated when the Month that gave it
birth comes again.

Your sbcl release for this month, sbcl-1.5.7, is now tagged in version
control and available for download from your most available SourceForge
mirror.  A list of changes relative to the previous version is appended
at the end of this message.

There are also some more details available about the upcoming 20th
birthday workshop in Vienna in December.  There is a web page for the
workshop at http://www.sbcl.org/sbcl20/ -- more information will be
added there, and most of the discussion about arrangements will be on
sbcl-devel -- I probably won't announce very much more here, so to be
included in further information please register at
https://forms.gle/FfHF8W8ptyPfMJir8 and subscribe, however temporarily,
to sbcl-devel.

Best wishes,

Christophe

changes in sbcl-1.5.7 relative to sbcl-1.5.6:
  * platform support:
    ** many bug fixes to the experimental 64-bit PowerPC/Linux port, to the
       extent that the little-endian variant passes all applicable tests in
       the regression test suite; the big-endian variant currently has some
       failures
    ** experimental support for sb-threads on 64-bit PowerPC/Linux
    ** support threads on x86-64 Sun OS (lp#1841280)
    ** handle PAX restrictions on mprotect() on NetBSD
    ** experimental support for HaikuOS
    ** the runtime is built as a position-independent executable by default on
       x86-64 Linux and x86-64 Darwin
  * fixes and enhancements related to Unicode:
    ** update of the data files to Unicode 8.0
    ** fix a bug in the implementation of the Unicode line breaking algorithm
       regarding hebrew letters and hyphens
  * enhancement: add a restart to OPEN with :IF-EXISTS :ERROR to allow
    re-opening with :APPEND.  (lp#806398, reported by Tobias Rittweiler)
  * bug fix: compiler optimizations on SEARCH with :FROM-END T didn't account
    for the empty sequence.  (lp#1844821)
  * bug fix: handle SETF of nested empty VALUES correctly.  (lp#1806478)
  * optimization: FLOOR and CEILING on rationals are simpler, and the
    compiler's understanding of them is better.