3 * patches postponed until after 0.7.0:
4 ** CSR "rough patch to fix bug 106" 2001-10-28
5 ** Alexey Dejneka "bug 111" 2001-12-30
6 * building with CLISP (or explaining why not). This will likely involve
7 a rearrangement of the build system so that it never renames
8 the output from COMPILE-FILE, because CLISP's COMPILE-FILE
9 outputs two (!) files and as far as I can tell LOAD uses both
10 of them. Since I have other motivations for this rearrangement
11 besides CLISPiosyncrasies, I'm reasonably motivated to do it.
12 * urgent EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
13 ** made inlining DEFUN inside MACROLET work again
14 ** (also, while working on INLINE anyway, it should be easy
15 to flush the old MAYBE-INLINE cruft entirely,
16 including e.g. on the man page)
17 ** fixed bug 137 (more)
18 * faster bootstrapping (both make.sh and slam.sh)
19 ** added mechanisms for automatically finding dead code, and
20 used them to remove dead code
21 ** moved stuff from warm init into cold init where possible
22 (so that slam.sh will run faster and also just because
23 ideally everything would be in cold init)
24 ** profiled and tweaked
25 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
26 ** made %COMPILE understand magicality of DEFUN FOO
27 w.r.t. e.g. preexisting inlineness of FOO
28 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
29 ** removed now-redundant COMPILE-TOP-LEVEL and
30 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
31 ** (ideally, but perhaps too hard, given what I've discovered
32 about the godawful internals of function debug names):
33 made FUNCTION-NAME logic work on closures, so that
34 various public functions like CL:PACKAGEP which
35 are now implemented as closures (because
36 they're structure slot accessors) won't be so
38 ** %SLOT-ACCESSOR/%SLOT-ACCESSOR stuff can probably go away,
39 since we inline expand all slot accessors into
40 %INSTANCE-REF and the optimizer knows all it needs
42 * rewrote long-standing confusing error restarts for redefining
44 * outstanding embarrassments
45 ** cut-and-pasted DEF-BOOLEAN-ATTRIBUTE (maybe easier to fix
46 now that EVAL-WHEN does what it should..)
48 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr. (It's
49 reasonable to support this as a crutch when initially
50 bootstrapping from balky xc hosts with their own
51 idiosyncratic ideas of what merits FAILURE-P, but it's
52 embarrassing to have to use it when bootstrapping
54 ** weird double-loading (first in GENESIS, then in warm init)
55 of src/assembly/target/*.lisp stuff, and the associated
56 weirdness of the half-baked state (compiler almost but
57 not quite ready for prime time..) of the system after
59 * fixups now feasible because of pre7 changes
60 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE)
61 * miscellaneous simple refactoring
63 ** renamed %PRIMITIVE to %VOP
64 * These days ANSI C has inline functions, so..
65 ** redid many cpp macros as inline functions:
66 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
67 GET_FREE_POINTER, SET_FREE_POINTER,
68 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
69 os_trunc_foo(), os_round_up_foo()
70 ** removed various avoid-evaluating-C-macro-arg-twice
72 * added mechanisms for automatically finding dead symbols is
73 package-data.lisp-expr (i.e. those symbols not bound,
74 fbound, defined as types, or whatever), and used them
75 to remove dead symbols
76 * made system handle stack overflow safely unless SAFETY is dominated
78 * Either get rid of or at least rework the fdefinition/encapsulation
79 system so that (SYMBOL-FUNCTION 'FOO) is identically equal to
81 =======================================================================
84 * refactored in preparation for moving CLOS into cold init and merging
85 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
86 ** systematized support for MOP (new regression tests, maybe
87 new SB-MOP package..) to try to make sure things don't
88 get mislaid in the upcoming CLOS restructuring
89 ** extracted type system from SB-KERNEL into new SB-TYPE
91 ** reimplemented GENERIC-FUNCTION as a primitive object (or
92 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
93 primitive object, and then let GENERIC-FUNCTIONs
94 inherit from that) instead of structures with
95 :ALTERNATE-METACLASS and funcallableness. Now
96 FUNCALLABLE-INSTANCE can go away. (And now the new
97 funcallable primitive objects need to go into
98 collections like *FUN-HEADER-WIDETAGS* where
99 FUNCALLABLE-INSTANCE objects used to be.)
100 ** reimplemented CONDITIONs as primitive objects instead of
101 structures with :ALTERNATE-METACLASS. Now (between
102 this and the change to GENERIC-FUNCTIONs)
103 DEFSTRUCT :ALTERNATE-METACLASS can go away.
104 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
105 STRUCTURE_POINTER_LOWTAG, and the concept of
106 SB-KERNEL:INSTANCE (including INSTANCEP,
107 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
108 * moved CLOS into cold init, in order to allow CLOS to be used in the
109 implementation of the core system (e.g. the type system and the
110 compiler) and in order to support merger of CL:CLASS with
112 * (maybe) eliminated warm init altogether in favor of cold init
113 * (maybe, especially if warm init can be eliminated) rationalized
114 the build process, fixing miscellaneous pre-0.5.0 stuff that's
115 transparently not the right thing
116 ** removed separate build directories, now just building in
117 place with .sbclcoldfasl extensions
118 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
119 into CL:FOO: reimplemented type system OO dispatch
120 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
121 * merged SB-PCL:FOO into CL:FOO (and similarly CLASS-OF, etc.)
122 * added some automatic tests for basic binary compatibility, in hopes
123 that it might be practical to maintain binary compatibility
124 between minor maintenance releases on the stable branch (but no
125 promises, sorry, since I've never tried to do this before, and
126 have no idea how much of a pain this'll be)
127 ========================================================================
128 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
129 to leave unfixed in 1.0):
130 * all too many BUGS entries and FIXMEs
131 =======================================================================
132 other priorities, no particular time:
134 * bug fixes, especially really annoying bugs (ANSI or not) and any
135 ANSI bugs (i.e. not just bugs in extras like the debugger or
136 "declarations are assertions", but violations of the standard)
137 * better communication with the outside world (scratching WHN's
138 personal itch): I don't want socket-level stuff so much as I
139 want RPC-level or higher (CORBA?) interfaces and (possibly
140 through RPC or CORBA) GUI support
141 =======================================================================
142 important but out of scope (for WHN, anyway: Patches from other people
143 are still welcome!) until after 1.0:
145 * sadly deteriorated support for ANSI-style block compilation
146 (static linking of DEFUNs within a single file or
147 WITH-COMPILATION-UNIT)
148 * various GC issues (exuberant cut-and-paste coding,
149 possibly dangerously over-conservative handling
150 of neighbors of function objects, general GC efficiency)
151 * package issues other than SB!TYPE, SB!MOP, and dead exported
153 * Any systematic effort to fix compiler consistency checks is
154 out of scope. (However, it still might be possible to
155 determine that some or all of them are hopelessly stale
157 ===============================================================================
158 other known issues with no particular target date:
160 bugs listed on the man page
162 hundreds of FIXME notes in the sources from WHN
164 various other unfinished business from CMU CL and before, marked with
165 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
166 or probably also other codes that I haven't noticed or have forgotten.
168 (Things marked as KLUDGE are in general things which are ugly or
169 confusing, but that, for whatever reason, may stay that way
171 =======================================================================
172 "There's nothing an agnostic can't do as long as he doesn't know
173 whether he believes in anything or not."
176 "God grant me serenity to accept the code I cannot change, courage to
177 change the code I can, and wisdom to know the difference."
180 "Accumulation of half-understood design decisions eventually chokes a
181 program as a water weed chokes a canal. By refactoring you can ensure
182 that your full understanding of how the program should be designed is
183 always reflected in the program. As a water weed quickly spreads its
184 tendrils, partially understood design decisions quickly spread their
185 effects throughout your program. No one or two or even ten individual
186 actions will be enough to eradicate the problem."
187 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
190 "I wish I didn't know now what I didn't know then."