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)
9 ** non-x86 ports now pass irrat.pure.lisp
10 ** sparc and ppc now pass bit-vector.impure-cload.lisp
11 * faster bootstrapping (both make.sh and slam.sh)
12 ** added mechanisms for automatically finding dead code, and
13 used them to remove dead code
14 ** moved stuff from warm init into cold init where possible
15 (so that slam.sh will run faster and also just because
16 ideally everything would be in cold init)
17 ** profiled and tweaked
18 * fixed (TRACE :REPORT PROFILE ...) interface to profiling
19 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
20 ** made %COMPILE understand magicality of DEFUN FOO
21 w.r.t. e.g. preexisting inlineness of FOO
22 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
23 ** removed now-redundant COMPILE-TOP-LEVEL and
24 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
25 ** (ideally, but perhaps too hard, given what I've discovered
26 about the godawful internals of function debug names):
27 made FUNCTION-NAME logic work on closures, so that
28 various public functions like CL:PACKAGEP which
29 are now implemented as closures (because
30 they're structure slot accessors) won't be so
32 * rewrote long-standing confusing error restarts for redefining
34 * outstanding embarrassments
35 ** cut-and-pasted DEF-BOOLEAN-ATTRIBUTE (maybe easier to fix
36 now that EVAL-WHEN works correctly..)
37 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr. (It's
38 reasonable to support this as a crutch when initially
39 bootstrapping from balky xc hosts with their own
40 idiosyncratic ideas of what merits FAILURE-P, but it's
41 embarrassing to have to use it when bootstrapping
43 ** weird double-loading (first in GENESIS, then in warm init)
44 of src/assembly/target/*.lisp stuff, and the associated
45 weirdness of the half-baked state (compiler almost but
46 not quite ready for prime time..) of the system after
48 * fixups now feasible because of pre7 changes
49 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE)
50 * miscellaneous simple refactoring
52 ** renamed %PRIMITIVE to %VOP
53 * These days ANSI C has inline functions, so..
54 ** redid many cpp macros as inline functions:
55 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
56 GET_FREE_POINTER, SET_FREE_POINTER,
57 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
58 os_trunc_foo(), os_round_up_foo()
59 ** removed various avoid-evaluating-C-macro-arg-twice
61 * added mechanisms for automatically finding dead symbols in
62 package-data.lisp-expr (i.e. those symbols not bound,
63 fbound, defined as types, or whatever), and used them
64 to remove dead symbols
65 * Either get rid of or at least rework the fdefinition/encapsulation
66 system so that (SYMBOL-FUNCTION 'FOO) is identically equal to
68 * building using CLISP (since building under OpenMCL works, this is
69 reduced to "it would be nice" rather than "as proof of concept")
71 =======================================================================
74 * refactored in preparation for moving CLOS into cold init and merging
75 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
76 ** systematized support for MOP (new regression tests, maybe
77 new SB-MOP package..) to try to make sure things don't
78 get mislaid in the upcoming CLOS restructuring
79 ** extracted type system from SB-KERNEL into new SB-TYPE
81 ** reimplemented GENERIC-FUNCTION as a primitive object (or
82 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
83 primitive object, and then let GENERIC-FUNCTIONs
84 inherit from that) instead of structures with
85 :ALTERNATE-METACLASS and funcallableness. Now
86 FUNCALLABLE-INSTANCE can go away. (And now the new
87 funcallable primitive objects need to go into
88 collections like *FUN-HEADER-WIDETAGS* where
89 FUNCALLABLE-INSTANCE objects used to be.)
90 ** reimplemented CONDITIONs as primitive objects instead of
91 structures with :ALTERNATE-METACLASS. Now (between
92 this and the change to GENERIC-FUNCTIONs)
93 DEFSTRUCT :ALTERNATE-METACLASS can go away.
94 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
95 STRUCTURE_POINTER_LOWTAG, and the concept of
96 SB-KERNEL:INSTANCE (including INSTANCEP,
97 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
98 * moved CLOS into cold init, in order to allow CLOS to be used in the
99 implementation of the core system (e.g. the type system and the
100 compiler) and in order to support merger of CL:CLASS with
102 * (maybe) eliminated warm init altogether in favor of cold init
103 * (maybe, especially if warm init can be eliminated) rationalized
104 the build process, fixing miscellaneous pre-0.5.0 stuff that's
105 transparently not the right thing
106 ** removed separate build directories, now just building in
107 place with .sbclcoldfasl extensions
108 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
109 into CL:FOO: reimplemented type system OO dispatch
110 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
111 * merged SB-PCL:FOO into CL:FOO (and similarly CLASS-OF, etc.)
112 * added some automatic tests for basic binary compatibility, in hopes
113 that it might be practical to maintain binary compatibility
114 between minor maintenance releases on the stable branch (but no
115 promises, sorry, since I've never tried to do this before, and
116 have no idea how much of a pain this'll be)
117 ========================================================================
118 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
119 to leave unfixed in 1.0):
120 * all too many BUGS entries and FIXMEs
121 =======================================================================
122 other priorities, no particular time:
124 * bug fixes, especially really annoying bugs (ANSI or not) and any
125 ANSI bugs (i.e. not just bugs in extras like the debugger or
126 "declarations are assertions", but violations of the standard)
127 * better communication with the outside world (scratching WHN's
128 personal itch): I don't want socket-level stuff so much as I
129 want RPC-level or higher (CORBA?) interfaces and (possibly
130 through RPC or CORBA) GUI support
131 =======================================================================
132 important but out of scope (for WHN, anyway: Patches from other people
133 are still welcome!) until after 1.0:
135 * sadly deteriorated support for ANSI-style block compilation
136 (static linking of DEFUNs within a single file or
137 WITH-COMPILATION-UNIT)
138 * various GC issues (exuberant cut-and-paste coding,
139 possibly dangerously over-conservative handling
140 of neighbors of function objects, general GC efficiency)
141 * package issues other than SB!TYPE, SB!MOP, and dead exported
143 * Any systematic effort to fix compiler consistency checks is
144 out of scope. (However, it still might be possible to
145 determine that some or all of them are hopelessly stale
147 =======================================================================
148 other known issues with no particular target date:
150 bugs listed on the man page
152 hundreds of FIXME notes in the sources from WHN
154 various other unfinished business from CMU CL and before, marked with
155 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
156 or probably also other codes that I haven't noticed or have forgotten.
158 (Things marked as KLUDGE are in general things which are ugly or
159 confusing, but that, for whatever reason, may stay that way
161 =======================================================================
162 "There's nothing an agnostic can't do as long as he doesn't know
163 whether he believes in anything or not."
166 "God grant me serenity to accept the code I cannot change, courage to
167 change the code I can, and wisdom to know the difference."
170 "Accumulation of half-understood design decisions eventually chokes a
171 program as a water weed chokes a canal. By refactoring you can ensure
172 that your full understanding of how the program should be designed is
173 always reflected in the program. As a water weed quickly spreads its
174 tendrils, partially understood design decisions quickly spread their
175 effects throughout your program. No one or two or even ten individual
176 actions will be enough to eradicate the problem."
177 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
180 "I wish I didn't know now what I didn't know then."