1 ;;;; tags which are set during the build process and which end up in
2 ;;;; CL:*FEATURES* in the target SBCL, plus some comments about other
3 ;;;; CL:*FEATURES* tags which have special meaning to SBCL or which
4 ;;;; have a special conventional meaning
6 ;;;; Note that the preferred way to customize the features of a local
7 ;;;; build of SBCL is not to edit this file, but to tweak
8 ;;;; customize-target-features.lisp.
10 ;;;; This software is part of the SBCL system. See the README file for
11 ;;;; more information.
13 ;;;; This software is derived from the CMU CL system, which was
14 ;;;; written at Carnegie Mellon University and released into the
15 ;;;; public domain. The software is in the public domain and is
16 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
17 ;;;; files for more information.
21 ;; features present in all builds
26 ;; FIXME: Isn't there a :x3jsomething feature which we should set too?
31 ;; Douglas Thomas Crosher's conservative generational GC (the only one
32 ;; we currently support)
35 ;; We're running under a UNIX. This is sort of redundant, and it was also
36 ;; sort of redundant under CMU CL, which we inherited it from: neither SBCL
37 ;; nor CMU CL supports anything but UNIX (and "technically not UNIX"es
38 ;; such as *BSD and Linux). But someday, maybe we might, and in that case
39 ;; we'd presumably remove this, so its presence conveys the information
40 ;; that the system isn't one which follows such a change.
44 ;; features present in this particular build
47 ;; Setting this enables the compilation of documentation strings
48 ;; from the system sources into the target Lisp executable.
49 ;; Traditional Common Lisp folk will want this option set.
50 ;; I (WHN) made it optional because I came to Common Lisp from
51 ;; C++ through Scheme, so I'm accustomed to asking
52 ;; Emacs about things that I'm curious about instead of asking
53 ;; the executable I'm running.
56 ;; When this is set, EVAL is implemented as an "IR1 interpreter":
57 ;; code is compiled into the compiler's first internal representation,
58 ;; then the IR1 is interpreted. When this is not set, EVAL is implemented
59 ;; as a little bit of hackery wrapped around a call to COMPILE, i.e.
60 ;; the system becomes a "compiler-only implementation" of Common Lisp.
61 ;; As of sbcl-0.6.7, the compiler-only implementation is prototype code,
62 ;; and much less mature than the old IR1 interpreter. Thus, the safe
63 ;; thing is to leave :SB-INTERPRETER set. However, the compiler-only
64 ;; system is noticeably smaller, so you might want to omit
65 ;; :SB-INTERPRETER if you have a small machine.
67 ;; Probably, the compiler-only implementation will become more
68 ;; stable someday, and support for the IR1 interpreter will then be
69 ;; dropped. This will make the system smaller and easier to maintain
70 ;; not only because we no longer need to support the interpreter,
71 ;; but because code elsewhere in the system (the dumper, the debugger,
72 ;; etc.) no longer needs special cases for interpreted code.
75 ;; Do regression and other tests when building the system. You
76 ;; might or might not want this if you're not a developer,
77 ;; depending on how paranoid you are. You probably do want it if
78 ;; you are a developer.
81 ;; Setting this makes more debugging information available.
82 ;; If you aren't hacking or troubleshooting SBCL itself, you
83 ;; probably don't want this set.
85 ;; At least two varieties of debugging information are enabled by this
87 ;; * SBCL is compiled with a higher level of OPTIMIZE DEBUG, so that
88 ;; the debugger can tell more about the state of the system.
89 ;; * Various code to print debugging messages, and similar debugging code,
90 ;; is compiled only when this feature is present.
92 ;; Note that the extra information recorded by the compiler at
93 ;; this higher level of OPTIMIZE DEBUG includes the source location
94 ;; forms. In order for the debugger to use this information, it has to
95 ;; re-READ the source file. In an ordinary installation of SBCL, this
96 ;; re-READing may not work very well, for either of two reasons:
97 ;; * The sources aren't present on the system in the same location that
98 ;; they were on the system where SBCL was compiled.
99 ;; * SBCL is using the standard readtable, without the added hackage
100 ;; which allows it to handle things like target features.
101 ;; If you want to be able to use the extra debugging information,
102 ;; therefore, be sure to keep the sources around, and run with the
103 ;; readtable configured so that the system sources can be read.
106 ;; Enable extra debugging output in the assem.lisp assembler/scheduler
107 ;; code. (This is the feature which was called :DEBUG in the
108 ;; original CMU CL code.)
111 ;; Setting this makes SBCL more "fluid", i.e. more amenable to
112 ;; modification at runtime, by suppressing various INLINE declarations,
113 ;; compiler macro definitions, FREEZE-TYPE declarations; and by
114 ;; suppressing various burning-our-ships-behind-us actions after
115 ;; initialization is complete; and so forth. This tends to clobber the
116 ;; performance of the system, so unless you have some special need for
117 ;; this when hacking SBCL itself, you don't want this set.
120 ;; Enable code for collecting statistics on usage of various operations,
121 ;; useful for performance tuning of the SBCL system itself. This code
122 ;; is probably pretty stale (having not been tested since the fork from
123 ;; base CMU CL) but might nonetheless be a useful starting point for
124 ;; anyone who wants to collect such statistics in the future.
127 ;; Peter Van Eynde's increase-bulletproofness code
129 ;; This is not maintained or tested in current SBCL, but I haven't
130 ;; gone out of my way to remove or break it, either.
133 ; :high-security-support
135 ;; multiprocessing support
137 ;; This is not maintained or tested in current SBCL. I haven't gone out
138 ;; of my way to break it, but since it's derived from an old version of
139 ;; CMU CL where multiprocessing was pretty shaky, it's likely to be very
141 ;; :MP enables multiprocessing
142 ;; :MP-I486 is used, only within the multiprocessing code, to control
143 ;; what seems to control processor-version-specific code. It's
144 ;; probably for 486 or later, i.e. could be set as long as
145 ;; you know you're not running on a 386, but it doesn't seem
146 ;; to be documented anywhere, so that's just a guess.
150 ;; KLUDGE: used to suppress stale code related to floating point infinities.
151 ;; I intend to delete this code completely some day, since it was a pain
152 ;; for me to try to work with and since all benefits it provides are
153 ;; non-portable. Until I actually pull the trigger, though, I've left
154 ;; various stale code in place protected with #!-SB-INFINITIES.
157 ;; This affects the definition of a lot of things in bignum.lisp. It
158 ;; doesn't seem to be documented anywhere what systems it might apply to.
159 ;; It doesn't seem to be needed for X86 systems anyway.
162 ;; This is probably true for some processor types, but not X86. It affects
163 ;; a lot of floating point code.
164 ; :negative-zero-is-not-zero
166 ;; This is mentioned in cmu-user.tex, which says that it enables
167 ;; the compiler to reason about integer arithmetic. It also seems to
168 ;; control other fancy numeric reasoning, e.g. knowing the result type of
169 ;; a remainder calculation given the type of its inputs.
171 ;; CROSS-FLOAT-INFINITY-KLUDGE: The :PROPAGATE-FLOAT-TYPE and
172 ;; :PROPAGATE-FUN-TYPE features are problematic when building
173 ;; the cross-compiler itself. Their implementation depends on
174 ;; floating point infinities, which might not be supported in the
175 ;; cross-compilation host. In order to avoid this problem, while
176 ;; still supporting these features in the target Lisp compiler,
177 ;; we use the :WILL-PROPAGATE-FLOAT-TYPE feature when building
178 ;; the cross-compiler, and munge it into :PROPAGATE-FLOAT-TYPE
179 ;; only when building the target compiler; and similarly for
180 ;; :WILL-PROPAGATE-FUN-TYPE.
181 ;:will-propagate-float-type ; (becomes :PROPAGATE-FLOAT-TYPE)
183 ;; According to cmu-user.tex, this enables the compiler to infer result
184 ;; types for mathematical functions like SQRT, EXPT, and LOG, allowing
185 ;; it to e.g. eliminate the possibility that a complex result will be
186 ;; generated. This applies only to the target compiler, not the
187 ;; cross-compiler: see CROSS-FLOAT-INFINITY-KLUDGE.
188 ;:will-propagate-fun-type ; (becomes :PROPAGATE-FUN-TYPE)
190 ;; It's unclear to me what this does (but it was enabled in the code that I
191 ;; picked up from Peter Van Eynde). -- WHN 19990224
192 :constrain-float-type
194 ;; This is set in classic CMU CL, and presumably there it means
195 ;; that the floating point arithmetic implementation
196 ;; conforms to IEEE's standard. Here it definitely means that the
197 ;; floating point arithmetic implementation conforms to IEEE's standard.
198 ;; I (WHN 19990702) haven't tried to verify
199 ;; that it does conform, but it should at least mostly conform (because
200 ;; the underlying x86 hardware tries).
203 ;; This seems to be the pre-GENCGC garbage collector for CMU CL, which was
204 ;; AFAIK never supported for the X86.
207 ;; CMU CL had, and we inherited, code to support 80-bit LONG-FLOAT on the x86
208 ;; architecture. Nothing has been done to actively destroy the long float
209 ;; support, but it hasn't been thoroughly maintained, and needs at least
210 ;; some maintenance before it will work. (E.g. the LONG-FLOAT-only parts of
211 ;; genesis are still implemented in terms of unportable CMU CL functions
212 ;; which are not longer available at genesis time in SBCL.) A deeper
213 ;; problem is SBCL's bootstrap process implicitly assumes that the
214 ;; cross-compilation host will be able to make the same distinctions
215 ;; between floating point types that it does. This assumption is
216 ;; fundamentally sleazy, even though in practice it's unlikely to break down
217 ;; w.r.t. distinguishing SINGLE-FLOAT from DOUBLE-FLOAT; it's much more
218 ;; likely to break down w.r.t. distinguishing DOUBLE-FLOAT from LONG-FLOAT.
219 ;; Still it's likely to be quite doable to get LONG-FLOAT support working
220 ;; again, if anyone's sufficiently motivated.
224 ;; miscellaneous notes on other things which could have special significance
225 ;; in the *FEATURES* list
228 ;; notes on the :NIL and :IGNORE features:
230 ;; #+NIL is used to comment out forms. Occasionally #+IGNORE is used
231 ;; for this too. So don't use :NIL or :IGNORE as the names of features..
233 ;; notes on :SB-XC and :SB-XC-HOST features (which aren't controlled by this
234 ;; file, but are instead temporarily pushed onto *FEATURES* or
235 ;; *TARGET-FEATURES* during some phases of cross-compilation):
237 ;; :SB-XC-HOST stands for "cross-compilation host" and is in *FEATURES*
238 ;; during the first phase of cross-compilation bootstrapping, when the
239 ;; host Lisp is being used to compile the cross-compiler.
241 ;; :SB-XC stands for "cross compiler", and is in *FEATURES* during the second
242 ;; phase of cross-compilation bootstrapping, when the cross-compiler is
243 ;; being used to create the first target Lisp.
245 ;; notes on the :PROPAGATE-FLOAT-TYPE and :PROPAGATE-FUN-TYPE
246 ;; features: See the comments on CROSS-FLOAT-INFINITY-KLUDGE.
248 ;; notes on the :SB-ASSEMBLING feature (which isn't controlled by
251 ;; This is a flag for whether we're in the assembler. It's
252 ;; temporarily pushed onto the *FEATURES* list in the setup for
253 ;; the ASSEMBLE-FILE function. It would be a bad idea
254 ;; to use it as a name for a permanent feature.
256 ;; notes on local features (which are set automatically by the
257 ;; configuration script, and should not be set here unless you
258 ;; really, really know what you're doing):
260 ;; machine architecture features:
261 ;; :x86 ; any Intel 386 or better, or compatibles like the AMD K6 or K7
262 ;; (No others are supported by SBCL as of 0.6.7, but :alpha or
263 ;; :sparc support could be ported from CMU CL if anyone is
264 ;; sufficiently motivated to do so.)
265 ;; (CMU CL also had a :pentium feature, which affected the definition
266 ;; of some floating point vops. It was present but not enabled in the
267 ;; CMU CL code that SBCL is derived from, and is present but stale
268 ;; in SBCL as of 0.6.7.)
270 ;; operating system features:
271 ;; :linux = We're intended to run under some version of Linux.
272 ;; :bsd = We're intended to run under some version of BSD Unix. (This
273 ;; is not exclusive with the features which indicate which
274 ;; particular version of BSD we're intended to run under.)
275 ;; :freebsd = We're intended to run under FreeBSD.
276 ;; :openbsd = We're intended to run under FreeBSD.
277 ;; (No others are supported by SBCL as of 0.6.7, but :hpux or
278 ;; :solaris support could be ported from CMU CL if anyone is
279 ;; sufficiently motivated to do so.)