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 * more renaming in global external names:
8 ** reserved DO-FOO-style names for iteration macros
11 ** perhaps s/DEF-FROB/DEF/ or s/DEF-FROB/DEFINE/
12 * Perhaps rename "cold" stuff (e.g. SB-COLD and src/cold/) to "boot".
13 * pending patches and bug reports that go in (or else get handled
14 somehow, rejected/logged/whatever) before 0.7.0:
15 =======================================================================
18 * patches postponed until after 0.7.0:
19 ** CSR "rough patch to fix bug 106" 2001-10-28
20 ** Alexey Dejneka "bug 111" 2001-12-30
21 * building with CLISP (or explaining why not). This will likely involve
22 a rearrangement of the build system so that it never renames
23 the output from COMPILE-FILE, because CLISP's COMPILE-FILE
24 outputs two (!) files and as far as I can tell LOAD uses both
25 of them. Since I have other motivations for this rearrangement
26 besides CLISPiosyncrasies, I'm reasonably motivated to do it.
27 * urgent EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
28 ** made inlining DEFUN inside MACROLET work again
29 ** (also, while working on INLINE anyway, it should be easy
30 to flush the old MAYBE-INLINE cruft entirely,
31 including e.g. on the man page)
32 ** fixed bug 137 (more)
33 * faster bootstrapping (both make.sh and slam.sh)
34 ** added mechanisms for automatically finding dead code, and
35 used them to remove dead code
36 ** moved stuff from warm init into cold init where possible
37 (so that slam.sh will run faster and also just because
38 ideally everything would be in cold init)
39 ** profiled and tweaked
40 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
41 ** made %COMPILE understand magicality of DEFUN FOO
42 w.r.t. e.g. preexisting inlineness of FOO
43 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
44 ** removed now-redundant COMPILE-TOP-LEVEL and
45 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
46 ** made FUNCTION-NAME logic work on closures, so that
47 various public functions like CL:PACKAGEP which
48 are now implemented as closures (because
49 they're structure slot accessors) won't be so
51 ** %SLOT-ACCESSOR/%SLOT-ACCESSOR stuff can probably go away,
52 since we inline expand all slot accessors into
53 %INSTANCE-REF and the optimizer knows all it needs
55 * rewrote long-standing confusing error restarts for redefining
57 * outstanding embarrassments
58 ** cut-and-pasted DEF-BOOLEAN-ATTRIBUTE (maybe easier to fix
59 now that EVAL-WHEN does what it should..)
61 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr
62 ** weird double-loading (first in GENESIS, then in warm init)
63 of src/assembly/target/*.lisp stuff, and the associated
64 weirdness of the half-baked state (compiler almost but
65 not quite ready for prime time..) of the system after
67 * fixups now feasible because of pre7 changes
68 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE)
69 * miscellaneous simple refactoring
71 ** renamed %PRIMITIVE to %VOP
72 * These days ANSI C has inline functions, so..
73 ** redid many cpp macros as inline functions:
74 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
75 GET_FREE_POINTER, SET_FREE_POINTER,
76 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
77 os_trunc_foo(), os_round_up_foo()
78 ** removed various avoid-evaluating-C-macro-arg-twice
80 * added mechanisms for automatically finding dead symbols is
81 package-data.lisp-expr (i.e. those symbols not bound,
82 fbound, defined as types, or whatever), and used them
83 to remove dead symbols
84 * made system handle stack overflow safely unless SAFETY is dominated
86 * Probably get rid of or at least rework the fdefinition/encapsulation
87 system so that (SYMBOL-FUNCTION 'FOO) = (FDEFINITION 'FOO).
88 =======================================================================
91 * refactored in preparation for moving CLOS into cold init and merging
92 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
93 ** systematized support for MOP (new regression tests, maybe
94 new SB-MOP package..) to try to make sure things don't
95 get mislaid in the upcoming CLOS restructuring
96 ** extracted type system from SB-KERNEL into new SB-TYPE
98 ** reimplemented GENERIC-FUNCTION as a primitive object (or
99 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
100 primitive object, and then let GENERIC-FUNCTIONs
101 inherit from that) instead of structures with
102 :ALTERNATE-METACLASS and funcallableness. Now
103 FUNCALLABLE-INSTANCE can go away. (And now the new
104 funcallable primitive objects need to go into
105 collections like *FUN-HEADER-WIDETAGS* where
106 FUNCALLABLE-INSTANCE objects used to be.)
107 ** reimplemented CONDITIONs as primitive objects instead of
108 structures with :ALTERNATE-METACLASS. Now (between
109 this and the change to GENERIC-FUNCTIONs)
110 DEFSTRUCT :ALTERNATE-METACLASS can go away.
111 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
112 STRUCTURE_POINTER_LOWTAG, and the concept of
113 SB-KERNEL:INSTANCE (including INSTANCEP,
114 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
115 * moved CLOS into cold init, in order to allow CLOS to be used in the
116 implementation of the core system (e.g. the type system and the
117 compiler) and in order to support merger of CL:CLASS with
119 * (maybe) eliminated warm init altogether in favor of cold init
120 * (maybe, especially if warm init can be eliminated) rationalized
121 the build process, fixing miscellaneous pre-0.5.0 stuff that's
122 transparently not the right thing
123 ** removed separate build directories, now just building in
124 place with .sbclcoldfasl extensions
125 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
126 into CL:FOO: reimplemented type system OO dispatch
127 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
128 * merged SB-PCL:FOO into CL:FOO (and similarly CLASS-OF, etc.)
129 * added some automatic tests for basic binary compatibility, in hopes
130 that it might be practical to maintain binary compatibility
131 between minor maintenance releases on the stable branch (but no
132 promises, sorry, since I've never tried to do this before, and
133 have no idea how much of a pain this'll be)
134 ========================================================================
135 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
136 to leave unfixed in 1.0):
137 * all too many BUGS entries and FIXMEs
138 =======================================================================
139 other priorities, no particular time:
141 * bug fixes, especially really annoying bugs (ANSI or not) and any
142 ANSI bugs (i.e. not just bugs in extras like the debugger or
143 "declarations are assertions", but violations of the standard)
144 * better communication with the outside world (scratching WHN's
145 personal itch): I don't want socket-level stuff so much as I
146 want RPC-level or higher (CORBA?) interfaces and (possibly
147 through RPC or CORBA) GUI support
148 =======================================================================
149 important but out of scope (for WHN, anyway: Patches from other people
150 are still welcome!) until after 1.0:
152 * sadly deteriorated support for ANSI-style block compilation
153 (static linking of DEFUNs within a single file or
154 WITH-COMPILATION-UNIT)
155 * various GC issues (exuberant cut-and-paste coding,
156 possibly dangerously over-conservative handling
157 of neighbors of function objects, general GC efficiency)
158 * package issues other than SB!TYPE, SB!MOP, and dead exported
160 * Any systematic effort to fix compiler consistency checks is
161 out of scope. (However, it still might be possible to
162 determine that some or all of them are hopelessly stale
164 ===============================================================================
165 other known issues with no particular target date:
167 bugs listed on the man page
169 hundreds of FIXME notes in the sources from WHN
171 various other unfinished business from CMU CL and before, marked with
172 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
173 or probably also other codes that I haven't noticed or have forgotten.
175 (Things marked as KLUDGE are in general things which are ugly or
176 confusing, but that, for whatever reason, may stay that way
178 =======================================================================
179 "There's nothing an agnostic can't do as long as he doesn't know
180 whether he believes in anything or not."
183 "God grant me serenity to accept the code I cannot change, courage to
184 change the code I can, and wisdom to know the difference."
187 "Accumulation of half-understood design decisions eventually chokes a
188 program as a water weed chokes a canal. By refactoring you can ensure
189 that your full understanding of how the program should be designed is
190 always reflected in the program. As a water weed quickly spreads its
191 tendrils, partially understood design decisions quickly spread their
192 effects throughout your program. No one or two or even ten individual
193 actions will be enough to eradicate the problem."
194 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
197 "I wish I didn't know now what I didn't know then."