1 <?xml version="1.0" encoding="iso-8859-1"?>
2 <!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1//EN"
3 "http://www.oasis-open.org/docbook/xml/4.1/docbookx.dtd" [
4 <!ENTITY % myents SYSTEM "entities.inc">
8 <chapter id="intro"><title>Introduction</title>
10 <para>&SBCL; is a mostly-conforming implementation of the &ANSI;
11 &CommonLisp; standard. This manual focuses on behavior which is
12 specific to &SBCL;, not on behavior which is common to all
13 implementations of &ANSI; &CommonLisp;.</para>
15 <sect1 id="more-cl-info">
16 <title>Where To Go For More Information about &CommonLisp; in General</title>
18 <para>Regardless of your ability level, two very useful resources
19 for working with any implementation of
21 <ulink url="http://ilisp.cons.org"><application>ILISP</application></ulink>
22 package for <application>Emacs</application> and
23 <ulink url="http://www.harlequin.com/books/HyperSpec">the &CommonLisp;
24 HyperSpec</ulink>.</para>
26 <para>If you're not a programmer and you're trying to learn,
27 many introductory Lisp books are available. However, we don't have any
28 standout favorites. If you can't decide, try checking the Usenet
29 comp.lang.lisp FAQ for recent recommendations.</para>
31 <para>If you are an experienced programmer in other languages
32 but need to learn about Lisp, three books stand out.
34 <listitem><para><emphasis>ANSI Common Lisp</emphasis>, by Paul Graham,
35 will teach you about most of the language. (And later it might
36 also be worth checking out <emphasis>On Lisp</emphasis>, by the same
37 author.)</para></listitem>
38 <listitem><para><emphasis>Paradigms Of Artificial Intelligence
39 Programming</emphasis>, by Peter Norvig, also has some good information
40 on general &CommonLisp; programming, and many nontrivial examples.
41 Whether or not your work is AI, it's a very good book to look
44 Neither of the books above emphasizes CLOS, but
45 <emphasis>Object-Oriented Programming In Common Lisp</emphasis> by Sonya Keene
46 does. Even if you're very knowledgeable about object oriented
47 programming in the abstract, it's worth looking at this book
48 if you want to do any OO in &CommonLisp;. Some abstractions
49 in CLOS (especially multiple dispatch) go beyond anything
50 you'll see in most OO systems, and there are a number of lesser
51 differences as well. This book tends to help with the culture shock.
58 <sect1 id="where-more">
59 <title>Where To Go For More Information About &SBCL;</title>
61 <para>Before you read this user manual, you should probably read
64 <listitem><para>You should know how to program in &CommonLisp;.
65 If you don't already know how, you should probably read a
66 <link linkend="more-cl-info">book on it</link>.</para></listitem>
67 <listitem><para>The Unix <quote>man page</quote> for &SBCL; will tell you
68 how to start the &SBCL; environment, so you can get to the
69 classic <quote>hello, world</quote> level of knowledge. It's the file
70 called <filename>sbcl.1</filename> in the &SBCL; distribution. If &SBCL; is
71 installed on your system, you can read a formatted copy by
72 executing the command <command>man sbcl</command>.</para></listitem>
76 <para>Besides this user manual and the Unix man page, some
77 other &SBCL;-specific information is available:
79 <listitem><para>The <ulink url="http://sbcl.sourceforge.net/">
80 &SBCL; home page</ulink> has some general
81 information, plus links to mailing lists devoted to &SBCL;,
82 and to archives of these mailing lists.</para></listitem>
83 <listitem><para>Documentation for non-&ANSI; extensions for
84 various commands is available online from the &SBCL; executable
85 itself. The extensions for functions which have their own
86 command prompts (e.g. the debugger, and <function>inspect</function>)
87 are documented in text available by typing <userinput>help</userinput>
88 at their command prompts. The extensions for functions which
89 don't have their own command prompt (like <function>trace</function>
90 does) are described in their documentation strings,
91 unless your &SBCL; was compiled with an option not
92 to include documentation strings, in which case the doc strings
93 are only readable in the source code.</para></listitem>
94 <listitem><para>Some low-level information describing the
95 programming details of the conversion from &CMUCL; to &SBCL;
96 is available in the <filename>doc/FOR-CMUCL-DEVELOPERS</filename>
97 file in the &SBCL; distribution.</para></listitem>
103 <sect1 id="implementation">
104 <title>Overview Of SBCL, How It Works And Where It Came From</title>
106 <para>You can work productively with SBCL without knowing anything
107 understanding anything about where it came from, how it is implemented,
108 or how it extends the &ANSI; &CommonLisp; standard. However,
109 a little knowledge can be helpful in order to understand error
110 messages, to troubleshoot problems, to understand why some parts of
111 the system are better debugged than others, and to anticipate which
112 known bugs, known performance problems, and missing extensions are
113 likely to be fixed, tuned, or added. </para>
115 <para>&SBCL; is descended from &CMUCL;, which is itself descended from
116 Spice Lisp, including early implementations for the Mach operating
117 system on the IBM RT, back in the 1980s. Design decisions from that
118 time are still reflected in the current implementation:
120 <listitem><para>The system expects to be loaded into a
121 fixed-at-compile-time location in virtual memory, and also expects
122 the location of all of its heap storage to be specified
123 at compile time.</para></listitem>
124 <listitem><para>The system overcommits memory, allocating large
125 amounts of address space from the system (often more than
126 the amount of virtual memory available) and then failing
127 if ends up using too much of the allocated storage.</para></listitem>
128 <listitem><para>A word is a 32-bit quantity. The system has been
129 ported to many processor architectures without altering this
130 basic principle. Some hacks allow the system to run on the Alpha
131 chip (a 64-bit architecture) but even there 32-bit words are
132 used. The assumption that a word is
133 32 bits wide is implicit in hundreds of places in the
134 system.</para></listitem>
135 <listitem><para>The system is implemented as a C program which is
136 responsible for supplying low-level services and loading a
137 Lisp <quote>.core</quote> file.
142 <para>&SBCL; also inherited some newer architectural features from
143 &CMUCL;. The most important is that it has a generational garbage
144 collector (<quote>GC</quote>), which has various implications (mostly good)
145 for performance. These are discussed in <link linkend="efficiency">
146 another chapter</link>.</para>
148 <para>&SBCL; has diverged from &CMUCL; in that &SBCL; is now
149 essentially a <quote>compiler-only implementation</quote> of
150 &CommonLisp;. A &CommonLisp; implementation is permitted to implement
151 both a compiler and an interpreter, and there's some special support
152 in the standard (e.g. the distinction between <function>functionp</function>
153 and <function>compiled-function-p</function>) to help support that. But &SBCL;
154 has only a vestigial, rudimentary true interpreter. In &SBCL;, the
155 <function>eval</function> function only truly <quote>interprets</quote> a few
156 special classes of forms, such as symbols which are
157 <function>boundp</function>. More complicated forms are evaluated by calling
158 <function>compile</function> and then calling <function>funcall</function> on the
162 <para>The direct ancestor of &SBCL; is the X86 port of &CMUCL;. This
163 port was in some ways the most cobbled-together of all the &CMUCL;
164 ports, since a number of strange changes had to be made to support the
165 register-poor X86 architecture. Some things (like tracing and
166 debugging) do not work particularly well there. &SBCL; should be able
167 to improve in these areas (and has already improved in some other
168 areas), but it takes a while.</para>
170 <para>On the x86, &SBCL; like the X86 port of &CMUCL;, uses a
171 <emphasis>conservative</emphasis> GC. This means that it doesn't maintain a
172 strict separation between tagged and untagged data, instead treating
173 some untagged data (e.g. raw floating point numbers) as
174 possibly-tagged data and so not collecting any Lisp objects that they
175 point to. This has some negative consequences for average time
176 efficiency (though possibly no worse than the negative consequences of
177 trying to implement an exact GC on a processor architecture as
178 register-poor as the X86) and also has potentially unlimited
179 consequences for worst-case memory efficiency. In practice,
180 conservative garbage collectors work reasonably well, not getting
181 anywhere near the worst case. But they can occasionally cause
182 odd patterns of memory usage.</para>
184 <para>The fork from &CMUCL; was based on a major rewrite of the system
185 bootstrap process. &CMUCL; has for many years tolerated a very unusual
186 <quote>build</quote> procedure which doesn't actually build the complete
187 system from scratch, but instead progressively overwrites parts of a
188 running system with new versions. This quasi-build procedure can cause
189 various bizarre bootstrapping hangups, especially when a major change
190 is made to the system. It also makes the connection between the
191 current source code and the current executable more tenuous than in
192 other software systems -- it's easy to accidentally
193 <quote>build</quote> a &CMUCL; system containing characteristics not
194 reflected in the current version of the source code.</para>
196 <para>Other major changes since the fork from &CMUCL; include
198 <listitem><para>&SBCL; has dropped support for many &CMUCL; extensions,
199 (e.g. IP networking, remote procedure call, Unix system interface, and X11
200 interface). Some of these are now available as contributed or
201 third-party modules.</para></listitem>
202 <listitem><para>&SBCL; has deleted or deprecated
203 some nonstandard features and code complexity which helped
204 efficiency at the price of maintainability. For example, the
205 &SBCL; compiler no longer implements memory pooling internally
206 (and so is simpler and more maintainable, but generates more
207 garbage and runs more slowly), and various block-compilation
208 efficiency-increasing extensions to the language have been
209 deleted or are no longer used in the implementation of &SBCL;
210 itself.</para></listitem>