4 ** non-x86 ports now pass irrat.pure.lisp
5 ** ports with less than 256Mb of heap (sparc, ppc and mips)
6 now don't fail bit-vector.impure-cload.lisp
7 * faster bootstrapping (both make.sh and slam.sh)
8 ** added mechanisms for automatically finding dead code, and
9 used them to remove dead code
10 ** moved stuff from warm init into cold init where possible
11 (so that slam.sh will run faster and also just because
12 ideally everything would be in cold init)
13 ** profiled and tweaked
14 * fixed (TRACE :REPORT PROFILE ...) interface to profiling
15 * more EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanup:
16 ** made %COMPILE understand magicality of DEFUN FOO
17 w.r.t. e.g. preexisting inlineness of FOO
18 ** used %COMPILE where COMPILE-TOP-LEVEL used to be used
19 ** removed now-redundant COMPILE-TOP-LEVEL and
20 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
21 ** (ideally, but perhaps too hard, given what I've discovered
22 about the godawful internals of function debug names):
23 made FUNCTION-NAME logic work on closures, so that
24 various public functions like CL:PACKAGEP which
25 are now implemented as closures (because
26 they're structure slot accessors) won't be so
28 * outstanding embarrassments
29 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr. (It's
30 reasonable to support this as a crutch when initially
31 bootstrapping from balky xc hosts with their own
32 idiosyncratic ideas of what merits FAILURE-P, but it's
33 embarrassing to have to use it when bootstrapping
35 * fixups now feasible because of pre7 changes
36 ** ANSIfied DECLAIM INLINE stuff (deprecating MAYBE-INLINE,
37 including e.g. on the man page)
38 ** (maybe) allow INLINE of a recursive function, so that the
40 * miscellaneous simple refactoring
42 ** renamed %PRIMITIVE to %VOP
43 ** A few hundred things named FN and FCN should be
44 named FUN (but maybe not while drichards is
45 working on a Windows port).
46 * These days ANSI C has inline functions, so..
47 ** redid many cpp macros as inline functions:
48 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
49 GET_FREE_POINTER, SET_FREE_POINTER,
50 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
51 os_trunc_foo(), os_round_up_foo()
52 ** removed various avoid-evaluating-C-macro-arg-twice
54 * Either get rid of or at least rework the fdefinition/encapsulation
55 system so that (SYMBOL-FUNCTION 'FOO) is identically equal to
57 * Make the system sources understandable to the system, so that
58 searching for sources doesn't error out quite so often
59 (e.g. in error handlers)
60 ** provided a location-independent way of referring to source
61 files in the target image, maybe a SYS: logical
62 pathname, and made the build system respect this.
63 ** provided a suitable readtable for reading in the source
64 files when necessary, and a mechanism for activating
65 this readtable rather than the standard one.
67 =======================================================================
70 [ note: much of the below refers to preparation for merging SB-PCL:FOO
71 and CL:FOO. However, it turned out to be surprisingly
72 straightforward to do this notional end goal without doing many of
73 the preparatory operations. That doesn't mean that plenty of the
74 goals below aren't worthwhile, but the motivation is somewhat
77 * refactored in preparation for moving CLOS into cold init and merging
78 SB-PCL:FOO with CL:FOO (for FOO=CLASS, FOO=CLASS-OF, etc.)
79 ** systematized support for MOP (more regression tests, maybe)
80 to try to make sure things don't get mislaid in the
81 upcoming CLOS restructuring
82 ** extracted type system (and maybe CLASSOIDs) from SB-KERNEL
83 into new SB-TYPE package
84 ** reimplemented GENERIC-FUNCTION as a primitive object (or
85 maybe made SB-MOP:FUNCALLABLE-STANDARD-OBJECT the
86 primitive object, and then let GENERIC-FUNCTIONs
87 inherit from that) instead of structures with
88 :ALTERNATE-METACLASS and funcallableness. Now
89 FUNCALLABLE-INSTANCE can go away. (And now the new
90 funcallable primitive objects need to go into
91 collections like *FUN-HEADER-WIDETAGS* where
92 FUNCALLABLE-INSTANCE objects used to be.)
93 ** reimplemented CONDITIONs as primitive objects instead of
94 structures with :ALTERNATE-METACLASS. Now (between
95 this and the change to GENERIC-FUNCTIONs)
96 DEFSTRUCT :ALTERNATE-METACLASS can go away.
97 ** (maybe) Now INSTANCE_POINTER_LOWTAG can become just
98 STRUCTURE_POINTER_LOWTAG, and the concept of
99 SB-KERNEL:INSTANCE (including INSTANCEP,
100 (SPECIFIER-TYPE 'INSTANCE), etc.) can go away.
101 * moved CLOS into cold init, in order to allow CLOS to be used in the
102 implementation of the core system (e.g. the type system and the
103 compiler) and in order to support merger of CL:CLASS with
105 * (maybe) eliminated warm init altogether in favor of cold init
106 * (maybe, especially if warm init can be eliminated) rationalized
107 the build process, fixing miscellaneous pre-0.5.0 stuff that's
108 transparently not the right thing
109 ** removed separate build directories, now just building in
110 place with .sbclcoldfasl extensions
111 * (maybe) more refactoring in preparation for merging SB-PCL:FOO
112 into CL:FOO: reimplemented type system OO dispatch
113 (!DEFINE-TYPE-METHOD, etc.) in terms of CLOS OO dispatch
114 * added some automatic tests for basic binary compatibility, in hopes
115 that it might be practical to maintain binary compatibility
116 between minor maintenance releases on the stable branch (but no
117 promises, sorry, since I've never tried to do this before, and
118 have no idea how much of a pain this'll be)
119 ========================================================================
120 for 1.0 (fixes of lower priority which I'd nonetheless be embarrassed
121 to leave unfixed in 1.0):
122 * all too many BUGS entries and FIXMEs
123 =======================================================================
124 other priorities, no particular time:
126 * bug fixes, especially really annoying bugs (ANSI or not) and any
127 ANSI bugs (i.e. not just bugs in extras like the debugger or
128 "declarations are assertions", but violations of the standard)
129 * better communication with the outside world (scratching WHN's
130 personal itch): I don't want socket-level stuff so much as I
131 want RPC-level or higher (CORBA?) interfaces and (possibly
132 through RPC or CORBA) GUI support
133 * Especially when ldb is not compiled in, the default "assertion failed"
134 behaviour in many parts of the runtime is unfriendly. It may
135 be appropriate to look at some of these and see if they can be
136 handled in some less abrupt way than aborting
137 =======================================================================
138 important but out of scope (for WHN, anyway: Patches from other people
139 are still welcome!) until after 1.0:
141 * sadly deteriorated support for ANSI-style block compilation
142 (static linking of DEFUNs within a single file or
143 WITH-COMPILATION-UNIT)
144 * various GC issues (exuberant cut-and-paste coding,
145 possibly dangerously over-conservative handling
146 of neighbors of function objects, general GC efficiency)
147 * package issues other than SB!TYPE, SB!MOP, and dead exported
149 * Any systematic effort to fix compiler consistency checks is
150 out of scope. (However, it still might be possible to
151 determine that some or all of them are hopelessly stale
153 =======================================================================
154 other known issues with no particular target date:
156 bugs listed on the man page
158 hundreds of FIXME notes in the sources from WHN
160 various other unfinished business from CMU CL and before, marked with
161 "XX" or "XXX" or "###" or "***" or "???" or "pfw" or "@@@@" or "zzzzz"
162 or probably also other codes that I haven't noticed or have forgotten.
164 (Things marked as KLUDGE are in general things which are ugly or
165 confusing, but that, for whatever reason, may stay that way
167 =======================================================================
168 "There's nothing an agnostic can't do as long as he doesn't know
169 whether he believes in anything or not."
172 "God grant me serenity to accept the code I cannot change, courage to
173 change the code I can, and wisdom to know the difference."
176 "Accumulation of half-understood design decisions eventually chokes a
177 program as a water weed chokes a canal. By refactoring you can ensure
178 that your full understanding of how the program should be designed is
179 always reflected in the program. As a water weed quickly spreads its
180 tendrils, partially understood design decisions quickly spread their
181 effects throughout your program. No one or two or even ten individual
182 actions will be enough to eradicate the problem."
183 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
186 "I wish I didn't know now what I didn't know then."