MNA patch for bug #30 and other readtable-related stuff
added tests for FIND and friends (anticipating new transforms)
fixed dumb error-reporting bug in CANONIZED-DECL-SPEC
fixed stupid wrong-stream bug in %DESCRIBE-COMPILED-FROM
MNA's filesys patches (from sbcl-devel 26 Dec 2000)
added DEFTRANSFORM for %WITH-ARRAY-DATA en route to speeding
up SORT, FIND, and other things
deleted *INITIAL-POLICY* and *INITIAL-INTERFACE-POLICY*, since
they seem irrelevant in the ANSI world of 1 source
file per compile command
deleted *INITIAL-PACKAGE* for the same reason
renamed *DEFAULT-POLICY* and *DEFAULT-INTERFACE-POLICY* to
*POLICY* and *INTERFACE-POLICY*
made PROCESS-OPTIMIZE-DECL remove duplicate alist entries
Happy New Year! May all your projects be on schedule.:-|
renamed POLICIES to LTN-POLICY
removed some inlining in ltn.lisp
simplified LTN-ANALYZE-BLOCK in ltn.lisp (removing caching)
added code to catch bogus full calls
tweaked PROFILE so it accepts (SETF FOO)-style names
tweaked DEFPRINTER to conserve whitespace
moved DEFPRINTER to SB-INT, since it's not compiler-specific
renamed more PCL stuff for unintern after warm init
CONSTANTLY now returns only a single value, as per ANSI.
removed CONSTANT-FUNCTION declaration support
PROCLAIM now recognizes ANSI abbreviated type declarations,
sharing code with DECLARE to do so.
DECLARE no longer supports old-style (CLTL1) FUNCTION decls.
removed some PCL nonstandard decls
renamed other nonstandard PCL decls to look more private
removed (DECLARE (SB-PCL::CLASS ..)) hack in DECLARE logic
(The "version broken" problem above was because the change
in POLICY type was not binary compatible, and I didn't
bump the version number, and I was crashing in some
private .sbclrc code compiled under the old system.
Oops..)
The POLICY-QUALITY type no longer includes NULL.
renamed more PCL stuff for unintern after warm init
(This version is broken. It builds, but can't build itself,
dying in vop.lisp with what looks like memory
corruption. 0.6.9.11 might be broken the same way.)
no more special support for DECLARE SB-PCL::CLASS
moved policy stuff into policy.lisp
moved !UNINTERN-INIT-ONLY-STUFF to after warm init
renamed some PCL stuff to make it be uninterned after warm init
more cleanups of optimization policy machinery..
renamed CSPEED slot to COMPILATION-SPEED, and BREVITY slot to
INHIBIT-WARNINGS, and got rid of other BREVITY refs
STRUCTURE-OBJECT-based POLICY caues too many cold init hassles
w/ COPY-POLICY and slot accessors. Use alists instead.
Now PROCESS-OPTIMIZE-DECLARATION can look up qualities
directly in *POLICY-BASIC-QUALITIES*.
ANSI fix: PARSE-NAMESTRING signals TYPE-ERROR on host mismatch.
various cleanups in optimization policy machinery..
renamed COOKIE to POLICY
used *POLICY-QUALITY-SLOTS* in POLICY-related def'ns
simplified POLICY macro: no implicit AND
factored out MAYBE-FP-WAIT policy dependence in float.lisp
added remaining MNA logical pathname regression tests to tests/
fixed my screwed-up CFP-OUTPUT-FILE-DEFAULT implementation
fixed PARSE-NAMESTRING so it always returns two values
added DEFKNOWNs for logical pathname stuff
MNA logical pathname patch (sbcl-devel Dec 12), lightly subbed
added most MNA logical pathname regression tests to tests/
belatedly bumped fasl file version because package
SB-CONDITIONS is gone
BACKEND-BYTE-FASL-FILE-TYPE goes away as per 0.6.9.4.
Simplify TRY-DEFAULT-TYPES and friends by using single default
types instead of lists of default types.
more work on bug #17, loosely based on MNA's patch; still broke
started tests/pathnames.impure.lisp
failed to fix cross-compilation under CMU CL
MNA software-version patch + Raymond Wiker FreeBSD corrections
started fix for bug #17 (loosely based on MNA's patch) by
ANSIfying COMPILE-FILE-PATHNAME
*DEFAULT-PATHNAME-DEFAULTS* and SEARCH-LIST stuff can be
initialized in a toplevel form, so !FILESYS-COLD-INIT
can go away.
added FLUSH-STANDARD-OUTPUT-STREAMS in INVOKE-DEBUGGER
applied Raymond Wiker's patch to share the stubs code from
Linux (instead of using BSD-only stubs)
redid DEFMETHOD macroexpansion, fixing bug 9 (problems with
IGNORE/IGNORABLE declarations) and generally
tidying up
added copyright boilerplate to tests/*.lisp
fixed bug 8 by replacing (ERROR 'PROGRAM-ERROR ..) by
(ERROR 'SIMPLE-PROGRAM-ERROR ..) everywhere
tweaked MNAME-SYM in DEFMETHOD to avoid dependence on *PACKAGE*
rewrote DEBUG-PROMPT to be less confusing when ILISP and SBCL
fight and end up 8+ levels deep in the debugger, then
relented and gave some advance warning of prompt change
deleted unsupported *DEBUG-PROMPT*
We don't really need an END-OF-THE-WORLD restart. The point was
to make sure a naive user can get out with Ctrl-C. So
mention (SB-EXT:QUIT) in entering-the-debugger message.
reduced choice-of-output-stream chaos in debugger
merged DTC's compound CONS type patches
enabled (and fixed) tests in tests/compound-cons.impure.lisp
canonicalized types (CONS *) etc. to (CONS T) etc.
removed "MNA: open-coded-simple-array patch"
Argh! typecase-implicit-declarations doesn't add declarations
after all.. just a messy no-op! And when I rewrite it
to 'work', I finally notice it's a fundamentally broken
idea. Gads. Delete it (and hope for real fix for #62).
left small rewrites in CASE-BODY: Naming quantities is good.
exported SB-KERNEL:%PUTHASH so DEFUN sees DEFKNOWN & vice versa
removed bogus FOLDABLEness from hash table accessors
SANE-PACKAGE now issues TYPE-ERROR, not just vanilla ERROR.
removed *COMPILE-TIME-DEFINE-MACROS*
added DTC comments clarifying that "MNA: dump-circular hack"
is a suboptimal solution
added underlying MNA typecase-implicit-declarations problem
to the BUGS list as bug #62
finished reviewing MNA megapatch (but haven't removed
rejected open-coded-simple-array-patch yet)
tested for bug 21, didn't find it, removed it from BUGS.
added SANE-PACKAGE to handle non-PACKAGE values of *PACKAGE*
deleted some unused and redundant stuff from PCL
tweaked DEFCONSTANTs to be more ANSI-compliant (as required
when building using an XC host incorporating changes
from the previous commit) and generally cleaner
got rid of DEFCONSTANT WRAPPER-LAYOUT completely, which was
used only by STRUCTURE-WRAPPER, which is now gone
added SB-INT:DEFCONSTANT-EQX to help ANSIfy DEFCONSTANTs
merged several small files into primordial-extensions.lisp
converted DEFMACRO DEFCONSTANT to use EVAL-WHEN instead of
IR1 magic, in order to make it ANSI enough for
DEFCONSTANT-EQX to work
removed various nested EVAL-WHENs (to help cross-compiler)
identified bug IR1-3, added workaround in DO-EVAL-WHEN-STUFF
incremented fasl file version (because of mismatch between
old IR1 magic %DEFCONSTANT/%%DEFCONSTANT behavior and
new EVAL-WHEN %DEFCONSTANT behavior)
deleted some unused code
fixed (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) bug
tweaked DEFCONSTANTs to be more ANSI-compliant (as required
when building using an XC host incorporating changes
from the previous commit) and generally cleaner
got rid of DEFCONSTANT WRAPPER-LAYOUT completely, which was
used only by STRUCTURE-WRAPPER, which is now gone
added SB-INT:DEFCONSTANT-EQX to help ANSIfy DEFCONSTANTs
merged several small files into primordial-extensions.lisp
converted DEFMACRO DEFCONSTANT to use EVAL-WHEN instead of
IR1 magic, in order to make it ANSI enough for
DEFCONSTANT-EQX to work
removed various nested EVAL-WHENs (to help cross-compiler)
identified bug IR1-3, added workaround in DO-EVAL-WHEN-STUFF
incremented fasl file version (because of mismatch between
old IR1 magic %DEFCONSTANT/%%DEFCONSTANT behavior and
new EVAL-WHEN %DEFCONSTANT behavior)
deleted some unused code
fixed (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) bug
