4 ** ports with less than 256Mb of heap (sparc, ppc and mips)
5 now don't fail bit-vector.impure-cload.lisp
6 * faster bootstrapping (both make.sh and slam.sh)
7 ** added mechanisms for automatically finding dead code, and
8 used them to remove dead code
9 ** moved stuff from warm init into cold init where possible
10 (so that slam.sh will run faster and also just because
11 ideally everything would be in cold init)
12 ** profiled and tweaked
13 * fixed (TRACE :REPORT PROFILE ...) interface to profiling
14 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
15 ** made %COMPILE understand magicality of DEFUN FOO
16 w.r.t. e.g. preexisting inlineness of FOO
17 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
18 ** removed now-redundant COMPILE-TOP-LEVEL and
19 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
20 ** (ideally, but perhaps too hard, given what I've discovered
21 about the godawful internals of function debug names):
22 made FUNCTION-NAME logic work on closures, so that
23 various public functions like CL:PACKAGEP which
24 are now implemented as closures (because
25 they're structure slot accessors) won't be so
27 * outstanding embarrassments
28 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr. (It's
29 reasonable to support this as a crutch when initially
30 bootstrapping from balky xc hosts with their own
31 idiosyncratic ideas of what merits FAILURE-P, but it's
32 embarrassing to have to use it when bootstrapping
34 * fixups now feasible because of pre7 changes
35 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE,
36 including e.g. on the man page)
37 ** (maybe) allow INLINE of a recursive function, so that the
39 * miscellaneous simple refactoring
41 ** renamed %PRIMITIVE to %VOP
42 ** A few hundred things named FN and FCN should be
43 named FUN (but maybe not while drichards is
44 working on a Windows port).
45 * These days ANSI C has inline functions, so..
46 ** redid many cpp macros as inline functions:
47 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
48 GET_FREE_POINTER, SET_FREE_POINTER,
49 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
50 os_trunc_foo(), os_round_up_foo()
51 ** removed various avoid-evaluating-C-macro-arg-twice
53 * Either get rid of or at least rework the fdefinition/encapsulation
54 system so that (SYMBOL-FUNCTION 'FOO) is identically equal to
56 * Make the system sources understandable to the system, so that
57 searching for sources doesn't error out quite so often
58 (e.g. in error handlers)
59 ** provided a suitable readtable for reading in the source
60 files when necessary, and a mechanism for activating
61 this readtable rather than the standard one.
62 * Some work on conditions emitted by the system
63 ** eliminated COMPILER-WARN and COMPILER-STYLE-WARN, which
64 were simply limited versions of WARN and STYLE-WARN.
65 ** made STYLE-WARN parallel WARN more closely (by accepting
66 a condition type, which should be a subtype of
67 STYLE-WARNING, and initargs, as well as a format
68 string and format arguments for SIMPLE-STYLE-WARNING.
69 (WARN can also be used to signal STYLE-WARNINGs, but
70 STYLE-WARN helps to document the code)
71 ** eliminated use of INHIBIT-WARNINGS by code emitted by the
72 system from user code.
73 ** caused use of INHIBIT-WARNINGS to signal a STYLE-WARNING.
74 ** eliminated use of INHIBIT-WARNINGS within the system
75 ** deprecated INHIBIT-WARNINGS, causing its use to signal a
77 ** began work on developing a class hierarchy of conditions
79 ** annotated conditions emitted by the system to have
80 references to documentation where applicable, so that
81 users can easily find an explanation for the
82 conditions they're seeing.
84 =======================================================================
87 [ note: much of the below refers to preparation for merging SB-PCL:FOO
88 and CL:FOO. However, it turned out to be surprisingly
89 straightforward to do this notional end goal without doing many of
90 the preparatory operations. That doesn't mean that plenty of the
91 goals below aren't worthwhile, but the motivation is somewhat
94 * refactored in preparation for moving CLOS into cold init and merging
95 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
96 ** systematized support for MOP (more regression tests, maybe)
97 to try to make sure things don't get mislaid in the
98 upcoming CLOS restructuring
99 ** extracted type system (and maybe CLASSOIDs) from SB-KERNEL
100 into new SB-TYPE package
101 ** reimplemented GENERIC-FUNCTION as a primitive object (or
102 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
103 primitive object, and then let GENERIC-FUNCTIONs
104 inherit from that) instead of structures with
105 :ALTERNATE-METACLASS and funcallableness. Now
106 FUNCALLABLE-INSTANCE can go away. (And now the new
107 funcallable primitive objects need to go into
108 collections like *FUN-HEADER-WIDETAGS* where
109 FUNCALLABLE-INSTANCE objects used to be.)
110 ** reimplemented CONDITIONs as primitive objects instead of
111 structures with :ALTERNATE-METACLASS. Now (between
112 this and the change to GENERIC-FUNCTIONs)
113 DEFSTRUCT :ALTERNATE-METACLASS can go away.
114 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
115 STRUCTURE_POINTER_LOWTAG, and the concept of
116 SB-KERNEL:INSTANCE (including INSTANCEP,
117 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
118 * moved CLOS into cold init, in order to allow CLOS to be used in the
119 implementation of the core system (e.g. the type system and the
120 compiler) and in order to support merger of CL:CLASS with
122 * (maybe) eliminated warm init altogether in favor of cold init
123 * (maybe, especially if warm init can be eliminated) rationalized
124 the build process, fixing miscellaneous pre-0.5.0 stuff that's
125 transparently not the right thing
126 ** removed separate build directories, now just building in
127 place with .sbclcoldfasl extensions
128 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
129 into CL:FOO: reimplemented type system OO dispatch
130 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
131 * added some automatic tests for basic binary compatibility, in hopes
132 that it might be practical to maintain binary compatibility
133 between minor maintenance releases on the stable branch (but no
134 promises, sorry, since I've never tried to do this before, and
135 have no idea how much of a pain this'll be)
136 ========================================================================
137 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
138 to leave unfixed in 1.0):
139 * all too many BUGS entries and FIXMEs
140 =======================================================================
141 other priorities, no particular time:
143 * bug fixes, especially really annoying bugs (ANSI or not) and any
144 ANSI bugs (i.e. not just bugs in extras like the debugger or
145 "declarations are assertions", but violations of the standard)
146 * better communication with the outside world (scratching WHN's
147 personal itch): I don't want socket-level stuff so much as I
148 want RPC-level or higher (CORBA?) interfaces and (possibly
149 through RPC or CORBA) GUI support
150 * Especially when ldb is not compiled in, the default "assertion failed"
151 behaviour in many parts of the runtime is unfriendly. It may
152 be appropriate to look at some of these and see if they can be
153 handled in some less abrupt way than aborting
154 =======================================================================
155 important but out of scope (for WHN, anyway: Patches from other people
156 are still welcome!) until after 1.0:
158 * sadly deteriorated support for ANSI-style block compilation
159 (static linking of DEFUNs within a single file or
160 WITH-COMPILATION-UNIT)
161 * various GC issues (exuberant cut-and-paste coding,
162 possibly dangerously over-conservative handling
163 of neighbors of function objects, general GC efficiency)
164 * package issues other than SB!TYPE, SB!MOP, and dead exported
166 * Any systematic effort to fix compiler consistency checks is
167 out of scope. (However, it still might be possible to
168 determine that some or all of them are hopelessly stale
170 =======================================================================
171 other known issues with no particular target date:
173 bugs listed on the man page
175 hundreds of FIXME notes in the sources from WHN
177 various other unfinished business from CMU CL and before, marked with
178 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
179 or probably also other codes that I haven't noticed or have forgotten.
181 (Things marked as KLUDGE are in general things which are ugly or
182 confusing, but that, for whatever reason, may stay that way
184 =======================================================================
185 "There's nothing an agnostic can't do as long as he doesn't know
186 whether he believes in anything or not."
189 "God grant me serenity to accept the code I cannot change, courage to
190 change the code I can, and wisdom to know the difference."
193 "Accumulation of half-understood design decisions eventually chokes a
194 program as a water weed chokes a canal. By refactoring you can ensure
195 that your full understanding of how the program should be designed is
196 always reflected in the program. As a water weed quickly spreads its
197 tendrils, partially understood design decisions quickly spread their
198 effects throughout your program. No one or two or even ten individual
199 actions will be enough to eradicate the problem."
200 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
203 "I wish I didn't know now what I didn't know then."