3 * filed off the roughest edges (or, perhaps, at least hammered down the
4 protruding rusty nails and snipped off the trailing razor wire,
5 leaving some filing for later:-) from the monster
6 EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
7 ** made inlining DEFUN inside MACROLET work again
8 * more renaming in global external names:
9 ** reserved DO-FOO-style names for iteration macros
10 ** finished s/FUNCTION/FUN/
12 ** s/TOP-LEVEL/TOPLEVEL/
13 * global style systematization:
14 ** s/#'(lambda/(lambda/
15 ** four-space indentation in C
16 * pending patches and bug reports that go in (or else get handled
17 somehow, rejected/logged/whatever) before 0.7.0:
18 ** AD "BUG in nested backquotes processing"
20 ** NJF bug report "bug in COPY-READTABLE" and AD patch,
21 both sbcl-devel 2001-12-24
22 ** AD patch for other readtable functions, sbcl-devel 2001-12-24
23 =======================================================================
26 * patches postponed until after 0.7.0:
27 ** Christophe Rhodes "rough patch to fix bug 106" 2001-10-28
28 * building with CLISP (or explaining why not)
29 * urgent EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
31 * faster bootstrapping (both make.sh and slam.sh)
32 ** added mechanisms for automatically finding dead code, and
33 used them to remove dead code
34 ** moved stuff from warm init into cold init where possible
35 (so that slam.sh will run faster and also just because
36 ideally everything would be in cold init)
37 ** profiled and tweaked
38 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
39 ** made %COMPILE understand magicality of DEFUN FOO
40 w.r.t. e.g. preexisting inlineness of FOO
41 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
42 ** removed now-redundant COMPILE-TOP-LEVEL and
43 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
44 ** made FUNCTION-NAME logic work on closures, so that
45 various public functions like CL:PACKAGEP which
46 are now implemented as closures (because
47 they're structure slot accessors) won't be so
49 ** %SLOT-ACCESSOR/%SLOT-ACCESSOR stuff can probably go away,
50 since we inline expand all slot accessors into
51 %INSTANCE-REF and the optimizer knows all it needs
53 * rewrote long-standing confusing error restarts for redefining
55 * outstanding embarrassments
56 ** cut-and-pasted DEF-BOOLEAN-ATTRIBUTE (maybe easier to fix
57 now that EVAL-WHEN does what it should..)
59 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr
60 ** weird double-loading (first in GENESIS, then in warm init)
61 of src/assembly/target/*.lisp stuff, and the associated
62 weirdness of the half-baked state (compiler almost but
63 not quite ready for prime time..) of the system after
65 * fixups now feasible because of pre7 changes
66 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE)
67 * miscellaneous simple refactoring
69 ** renamed %PRIMITIVE to %VOP
70 * These days ANSI C has inline functions, so..
71 ** redid many cpp macros as inline functions:
72 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
73 GET_FREE_POINTER, SET_FREE_POINTER,
74 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
75 os_trunc_foo(), os_round_up_foo()
76 ** removed various avoid-evaluating-C-macro-arg-twice
78 * added mechanisms for automatically finding dead symbols is
79 package-data.lisp-expr (i.e. those symbols not bound,
80 fbound, defined as types, or whatever), and used them
81 to remove dead symbols
82 * made system handle stack overflow safely unless SAFETY is dominated
84 * Probably get rid of or at least rework the fdefinition/encapsulation
85 system so that (SYMBOL-FUNCTION 'FOO) = (FDEFINITION 'FOO).
86 =======================================================================
89 * refactored in preparation for moving CLOS into cold init and merging
90 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
91 ** systematized support for MOP (new regression tests, maybe
92 new SB-MOP package..) to try to make sure things don't
93 get mislaid in the upcoming CLOS restructuring
94 ** extracted type system from SB-KERNEL into new SB-TYPE
96 ** reimplemented GENERIC-FUNCTION as a primitive object (or
97 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
98 primitive object, and then let GENERIC-FUNCTIONs
99 inherit from that) instead of structures with
100 :ALTERNATE-METACLASS and funcallableness. Now
101 FUNCALLABLE-INSTANCE can go away. (And now the new
102 funcallable primitive objects need to go into
103 collections like *FUN-HEADER-WIDETAGS* where
104 FUNCALLABLE-INSTANCE objects used to be.)
105 ** reimplemented CONDITIONs as primitive objects instead of
106 structures with :ALTERNATE-METACLASS. Now (between
107 this and the change to GENERIC-FUNCTIONs)
108 DEFSTRUCT :ALTERNATE-METACLASS can go away.
109 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
110 STRUCTURE_POINTER_LOWTAG, and the concept of
111 SB-KERNEL:INSTANCE (including INSTANCEP,
112 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
113 * moved CLOS into cold init, in order to allow CLOS to be used in the
114 implementation of the core system (e.g. the type system and the
115 compiler) and in order to support merger of CL:CLASS with
117 * (maybe) eliminated warm init altogether in favor of cold init
118 * (maybe, especially if warm init can be eliminated) rationalized
119 the build process, fixing miscellaneous pre-0.5.0 stuff that's
120 transparently not the right thing
121 ** removed separate build directories, now just building in
122 place with .sbclcoldfasl extensions
123 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
124 into CL:FOO: reimplemented type system OO dispatch
125 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
126 * merged SB-PCL:FOO into CL:FOO (and similarly CLASS-OF, etc.)
127 * added some automatic tests for basic binary compatibility, in hopes
128 that it might be practical to maintain binary compatibility
129 between minor maintenance releases on the stable branch (but no
130 promises, sorry, since I've never tried to do this before, and
131 have no idea how much of a pain this'll be)
132 ========================================================================
133 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
134 to leave unfixed in 1.0):
135 * all too many BUGS entries and FIXMEs
136 =======================================================================
137 other priorities, no particular time:
139 * bug fixes, especially really annoying bugs (ANSI or not) and any
140 ANSI bugs (i.e. not just bugs in extras like the debugger or
141 "declarations are assertions", but violations of the standard)
142 * better communication with the outside world (scratching WHN's
143 personal itch): I don't want socket-level stuff so much as I
144 want RPC-level or higher (CORBA?) interfaces and (possibly
145 through RPC or CORBA) GUI support
146 =======================================================================
147 important but out of scope (for WHN, anyway: Patches from other people
148 are still welcome!) until after 1.0:
150 * sadly deteriorated support for ANSI-style block compilation
151 (static linking of DEFUNs within a single file or
152 WITH-COMPILATION-UNIT)
153 * various GC issues (exuberant cut-and-paste coding,
154 possibly dangerously over-conservative handling
155 of neighbors of function objects, general GC efficiency)
156 * package issues other than SB!TYPE, SB!MOP, and dead exported
158 * Any systematic effort to fix compiler consistency checks is
159 out of scope. (However, it still might be possible to
160 determine that some or all of them are hopelessly stale
162 ===============================================================================
163 other known issues with no particular target date:
165 bugs listed on the man page
167 hundreds of FIXME notes in the sources from WHN
169 various other unfinished business from CMU CL and before, marked with
170 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
171 or probably also other codes that I haven't noticed or have forgotten.
173 (Things marked as KLUDGE are in general things which are ugly or
174 confusing, but that, for whatever reason, may stay that way
176 =======================================================================
177 "There's nothing an agnostic can't do as long as he doesn't know
178 whether he believes in anything or not."
181 "God grant me serenity to accept the code I cannot change, courage to
182 change the code I can, and wisdom to know the difference."
185 "Accumulation of half-understood design decisions eventually chokes a
186 program as a water weed chokes a canal. By refactoring you can ensure
187 that your full understanding of how the program should be designed is
188 always reflected in the program. As a water weed quickly spreads its
189 tendrils, partially understood design decisions quickly spread their
190 effects throughout your program. No one or two or even ten individual
191 actions will be enough to eradicate the problem."
192 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
195 "I wish I didn't know now what I didn't know then."