4 "There's nothing an agnostic can't do as long as he doesn't know
5 whether he believes in anything or not."
8 "God grant me serenity to accept the code I cannot change, courage
9 to change the code I can, and wisdom to know the difference."
12 "Accumulation of half-understood design decisions eventually
13 chokes a program as a water weed chokes a canal. By refactoring
14 you can ensure that your full understanding of how the program
15 should be designed is always reflected in the program. As a water
16 weed quickly spreads its tendrils, partially understood design
17 decisions quickly spread their effects throughout your program. No
18 one or two or even ten individual actions will be enough to
19 eradicate the problem."
20 -- Martin Fowler, in _Refactoring: Improving the Design of Existing
23 "I wish I didn't know now what I didn't know then."
26 This files is maintained as part of the SBCL distribution, and
27 describes flying pies and moons on sticks that SBCL developers dream
28 about. The items are in no particular order.
30 The omission of an item is no guarantee that no-one would like it,
31 just like the inclusion of an item is no guarantee that someone is
32 actively working on it.
34 In addition to this file, there is the BUGS file, and there are also
35 hundreds of FIXME notes in the sources. (Things marked as KLUDGE are
36 in general things which are ugly or confusing, but that, for
37 whatever reason, may stay that way indefinitely.)
41 SB-THREAD has some problems: recursivity of a mutex should probably
42 be a feature of the mutex, and not of the lexical location. Thread
43 local variables are needed. Sessions and sharing the terminal need
48 Have you ever read SBCL disassembly?
52 Global lexical variables. Esp. since with threads special variable
53 accesses is no speed daemon.
55 FINISHING EXTERNAL FORMATS
57 Byte order marks. Newline conventions. A way to specify an external
58 format without needing to duplicate code. Fixing the inefficiencies.
62 Asyncronous unwinds suck horribly, but to implement reliable systems
63 one simply needs timeouts. These should probably be local, since
64 otherwise they effectively become asynchronous unwinds. Basically,
65 for any potentially blocking operation there should be a :TIMEOUT
66 arguent (or a version of the operation that accepts the argument).
70 SBCL has an internal function encapsulation mechanism, and is able to
71 install breakpoint to function start/end -- this is used to implement
72 the instrumentation based profiler and tracing. It would be good to
73 have this as an exported interface, and it would be good if the
74 SYMBOL-FUNCTION / FDEFINITION confusion was fixed: currently the
75 latter returns the underlying definition, whereas the first returns
78 GENERIC FUNCTION TRACING
80 This sucks currently. It would also be good to be able to trace
85 The interactions between various optimization policies are far from
86 obvious. Someone should figure out how to make this better, and fix
87 it. One option would be to have just a few (eg. DEBUG, SMALL,
88 FAST-SAFE, FAST-UNSAFE) "dominant" policies, and expose the rest
89 as separately declarable optimization toggles.
91 MAYBE-INLINE is also nice, but it would be good if someone could
92 figure out how to get rid of it while retaining the semantics it
93 provides. Inlining recursive functions is also something to think
96 INHIBIT-WARNINGS really needs to go away.
104 Needs love, particularly threads and exceptions/signals. slam.sh is
108 Multithreading does not work. Possibly because of bug in mmap/munmap.
109 Hint: Comment out call to os_invalidate in perform_thread_post_mortem
110 and threads will work, but space will not be freed, of course.
112 UNDEFINED FUNCTION / VARIABLE RESTARTS
114 You know, like Allegro is reputed to have.
118 These need to be taken a good look at -- it might be that some of them
119 are already done, or otherwise no longer relevant.)
121 * EVAL/EVAL-WHEN/%COMPILE/DEFUN/DEFSTRUCT cleanups:
122 ** make %COMPILE understand magicality of DEFUN FOO
123 w.r.t. e.g. preexisting inlineness of FOO
124 ** use %COMPILE where COMPILE-TOP-LEVEL used to be used
125 ** remove redundant COMPILE-TOP-LEVEL and
126 FUNCTIONAL-KIND=:TOP-LEVEL stuff from the compiler
127 * outstanding embarrassments
128 ** :IGNORE-ERRORS-P cruft in stems-and-flags.lisp-expr. (It's
129 reasonable to support this as a crutch when initially
130 bootstrapping from balky xc hosts with their own
131 idiosyncratic ideas of what merits FAILURE-P, but it's
132 embarrassing to have to use it when bootstrapping
134 * miscellaneous simple refactoring
136 ** rename %PRIMITIVE to %VOP
137 ** A few hundred things named FN should be named FUN
138 * These days ANSI C has inline functions, so..
139 ** redo many cpp macros as inline functions:
140 HeaderValue, Pointerp, CEILING, ALIGNED_SIZE,
141 GET_FREE_POINTER, SET_FREE_POINTER,
142 GET_GC_TRIGGER, SET_GC_TRIGGER, GetBSP, SetBSP,
143 os_trunc_foo(), os_round_up_foo()
144 ** remove various avoid-evaluating-C-macro-arg-twice
146 * Some work on conditions emitted by the system
147 ** eliminate COMPILER-WARN and COMPILER-STYLE-WARN, which
148 were simply limited versions of WARN and STYLE-WARN.
149 ** eliminate use of INHIBIT-WARNINGS by code emitted by the
150 system from user code.
151 ** cause use of INHIBIT-WARNINGS to signal a STYLE-WARNING.
152 ** eliminate use of INHIBIT-WARNINGS within the system
153 ** deprecate INHIBIT-WARNINGS, causing its use to signal a
155 ** begin work on developing a class hierarchy of conditions
156 along semantic lines.
160 AKA CLOS in cold init.
162 HIGH LEVEL SOCKET INTERFACE
164 Something slightly above the level of BSD sockets would be nice.
168 For talking with other processes.
172 Since this is a high priority we're waiting for PHP coders to
173 offer their help to build a website about this.
177 You know, like CMUCL does.
181 Several algorithms exist, even one would be nice.
185 For those who need it.