3 * urgent EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
4 ** made inlining DEFUN inside MACROLET work again
5 ** (also, while working on INLINE anyway, it might be easy
6 to flush the old MAYBE-INLINE cruft entirely,
7 including e.g. on the man page)
8 ** fixed bug 137 (more)
9 * faster bootstrapping (both make.sh and slam.sh)
10 ** added mechanisms for automatically finding dead code, and
11 used them to remove dead code
12 ** moved stuff from warm init into cold init where possible
13 (so that slam.sh will run faster and also just because
14 ideally everything would be in cold init)
15 ** profiled and tweaked
16 * fixed (TRACE :REPORT PROFILE ...) interface to profiling
17 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
18 ** made %COMPILE understand magicality of DEFUN FOO
19 w.r.t. e.g. preexisting inlineness of FOO
20 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
21 ** removed now-redundant COMPILE-TOP-LEVEL and
22 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
23 ** (ideally, but perhaps too hard, given what I've discovered
24 about the godawful internals of function debug names):
25 made FUNCTION-NAME logic work on closures, so that
26 various public functions like CL:PACKAGEP which
27 are now implemented as closures (because
28 they're structure slot accessors) won't be so
30 * rewrote long-standing confusing error restarts for redefining
32 * outstanding embarrassments
33 ** cut-and-pasted DEF-BOOLEAN-ATTRIBUTE (maybe easier to fix
34 now that EVAL-WHEN works correctly..)
36 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr. (It's
37 reasonable to support this as a crutch when initially
38 bootstrapping from balky xc hosts with their own
39 idiosyncratic ideas of what merits FAILURE-P, but it's
40 embarrassing to have to use it when bootstrapping
42 ** weird double-loading (first in GENESIS, then in warm init)
43 of src/assembly/target/*.lisp stuff, and the associated
44 weirdness of the half-baked state (compiler almost but
45 not quite ready for prime time..) of the system after
47 * fixups now feasible because of pre7 changes
48 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE)
49 * miscellaneous simple refactoring
51 ** renamed %PRIMITIVE to %VOP
52 * These days ANSI C has inline functions, so..
53 ** redid many cpp macros as inline functions:
54 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
55 GET_FREE_POINTER, SET_FREE_POINTER,
56 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
57 os_trunc_foo(), os_round_up_foo()
58 ** removed various avoid-evaluating-C-macro-arg-twice
60 * added mechanisms for automatically finding dead symbols in
61 package-data.lisp-expr (i.e. those symbols not bound,
62 fbound, defined as types, or whatever), and used them
63 to remove dead symbols
64 * Either get rid of or at least rework the fdefinition/encapsulation
65 system so that (SYMBOL-FUNCTION 'FOO) is identically equal to
67 * building using CLISP (since building under OpenMCL works, this is
68 reduced to "it would be nice" rather than "as proof of concept")
70 =======================================================================
73 * refactored in preparation for moving CLOS into cold init and merging
74 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
75 ** systematized support for MOP (new regression tests, maybe
76 new SB-MOP package..) to try to make sure things don't
77 get mislaid in the upcoming CLOS restructuring
78 ** extracted type system from SB-KERNEL into new SB-TYPE
80 ** reimplemented GENERIC-FUNCTION as a primitive object (or
81 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
82 primitive object, and then let GENERIC-FUNCTIONs
83 inherit from that) instead of structures with
84 :ALTERNATE-METACLASS and funcallableness. Now
85 FUNCALLABLE-INSTANCE can go away. (And now the new
86 funcallable primitive objects need to go into
87 collections like *FUN-HEADER-WIDETAGS* where
88 FUNCALLABLE-INSTANCE objects used to be.)
89 ** reimplemented CONDITIONs as primitive objects instead of
90 structures with :ALTERNATE-METACLASS. Now (between
91 this and the change to GENERIC-FUNCTIONs)
92 DEFSTRUCT :ALTERNATE-METACLASS can go away.
93 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
94 STRUCTURE_POINTER_LOWTAG, and the concept of
95 SB-KERNEL:INSTANCE (including INSTANCEP,
96 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
97 * moved CLOS into cold init, in order to allow CLOS to be used in the
98 implementation of the core system (e.g. the type system and the
99 compiler) and in order to support merger of CL:CLASS with
101 * (maybe) eliminated warm init altogether in favor of cold init
102 * (maybe, especially if warm init can be eliminated) rationalized
103 the build process, fixing miscellaneous pre-0.5.0 stuff that's
104 transparently not the right thing
105 ** removed separate build directories, now just building in
106 place with .sbclcoldfasl extensions
107 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
108 into CL:FOO: reimplemented type system OO dispatch
109 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
110 * merged SB-PCL:FOO into CL:FOO (and similarly CLASS-OF, etc.)
111 * added some automatic tests for basic binary compatibility, in hopes
112 that it might be practical to maintain binary compatibility
113 between minor maintenance releases on the stable branch (but no
114 promises, sorry, since I've never tried to do this before, and
115 have no idea how much of a pain this'll be)
116 ========================================================================
117 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
118 to leave unfixed in 1.0):
119 * all too many BUGS entries and FIXMEs
120 =======================================================================
121 other priorities, no particular time:
123 * bug fixes, especially really annoying bugs (ANSI or not) and any
124 ANSI bugs (i.e. not just bugs in extras like the debugger or
125 "declarations are assertions", but violations of the standard)
126 * better communication with the outside world (scratching WHN's
127 personal itch): I don't want socket-level stuff so much as I
128 want RPC-level or higher (CORBA?) interfaces and (possibly
129 through RPC or CORBA) GUI support
130 =======================================================================
131 important but out of scope (for WHN, anyway: Patches from other people
132 are still welcome!) until after 1.0:
134 * sadly deteriorated support for ANSI-style block compilation
135 (static linking of DEFUNs within a single file or
136 WITH-COMPILATION-UNIT)
137 * various GC issues (exuberant cut-and-paste coding,
138 possibly dangerously over-conservative handling
139 of neighbors of function objects, general GC efficiency)
140 * package issues other than SB!TYPE, SB!MOP, and dead exported
142 * Any systematic effort to fix compiler consistency checks is
143 out of scope. (However, it still might be possible to
144 determine that some or all of them are hopelessly stale
146 =======================================================================
147 other known issues with no particular target date:
149 bugs listed on the man page
151 hundreds of FIXME notes in the sources from WHN
153 various other unfinished business from CMU CL and before, marked with
154 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
155 or probably also other codes that I haven't noticed or have forgotten.
157 (Things marked as KLUDGE are in general things which are ugly or
158 confusing, but that, for whatever reason, may stay that way
160 =======================================================================
161 "There's nothing an agnostic can't do as long as he doesn't know
162 whether he believes in anything or not."
165 "God grant me serenity to accept the code I cannot change, courage to
166 change the code I can, and wisdom to know the difference."
169 "Accumulation of half-understood design decisions eventually chokes a
170 program as a water weed chokes a canal. By refactoring you can ensure
171 that your full understanding of how the program should be designed is
172 always reflected in the program. As a water weed quickly spreads its
173 tendrils, partially understood design decisions quickly spread their
174 effects throughout your program. No one or two or even ten individual
175 actions will be enough to eradicate the problem."
176 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
179 "I wish I didn't know now what I didn't know then."