| blob | 5115bcb040990f1e7820cfc23af9e38ea6ca7d73 |
1 ;;;; needed-early, or at least meaningful-early, stuff for FASL files
3 ;;;; This software is part of the SBCL system. See the README file for
4 ;;;; more information.
5 ;;;;
6 ;;;; This software is derived from the CMU CL system, which was
7 ;;;; written at Carnegie Mellon University and released into the
8 ;;;; public domain. The software is in the public domain and is
9 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
10 ;;;; files for more information.
13 \f
14 ;;;; various constants and essentially-constants
16 ;;; a string which appears at the start of a fasl file header
17 ;;;
18 ;;; This value is used to identify fasl files. Even though this is not
19 ;;; declared as a constant (because ANSI Common Lisp has no facility
20 ;;; for declaring values which are constant under EQUAL but not EQL),
21 ;;; obviously you shouldn't mess with it lightly. If you do set a new
22 ;;; value for some reason, keep these things in mind:
23 ;;; * To avoid confusion with the similar but incompatible CMU CL
24 ;;; fasl file format, the value should not be "FASL FILE", which
25 ;;; is what CMU CL used for the same purpose.
26 ;;; * Since its presence at the head of a file is used by LOAD to
27 ;;; decide whether a file is to be fasloaded or just loaded
28 ;;; ordinarily (as source), the value should be something which
29 ;;; can't legally appear at the head of a Lisp source file.
30 ;;; * The value should not contain any line-terminating characters,
31 ;;; because they're hard to express portably and because the LOAD
32 ;;; code might reasonably use READ-LINE to get the value to compare
33 ;;; against.
40 ;; a list of *(SHEBANG-)FEATURES* flags which affect
41 ;; binary compatibility, i.e. which must be the same
42 ;; between the SBCL which compiles the code and the
43 ;; SBCL which executes the code
44 ;;
45 ;; This is a property of SBCL executables in the
46 ;; abstract, not of this particular SBCL executable,
47 ;; so any flag in this list may or may not be present
48 ;; in the *FEATURES* list of this particular build.
51 ;; a string representing flags of *F-P-A-F-F* which
52 ;; are in this particular build
53 ;;
54 ;; (A list is the natural logical representation for
55 ;; this, but we represent it as a string because
56 ;; that's physically convenient for writing to and
57 ;; reading from fasl files, and because we don't
58 ;; need to do anything sophisticated with its
59 ;; logical structure, just test it for equality.)
70 ;;; the code for a character which terminates a fasl file header
73 ;;; This value should be incremented when the system changes in such a
74 ;;; way that it will no longer work reliably with old fasl files. In
75 ;;; practice, I (WHN) have often forgotten to increment it for CVS
76 ;;; versions which break binary compatibility. But it certainly should
77 ;;; be incremented for release versions which break binary
78 ;;; compatibility.
80 ;;; (record of versions before 2003 deleted in 2003-04-26/0.pre8.107 or so)
81 ;;; 38: (2003-01-05) changed names of internal SORT machinery
82 ;;; 39: (2003-02-20) in 0.7.12.1 a slot was added to
83 ;;; DEFSTRUCT-SLOT-DESCRIPTION
84 ;;; 40: (2003-03-11) changed value of (SXHASH NIL)
85 ;;; 41: (2003-04-26) enforced binary incompatibility between +SB-THREAD
86 ;;; and -SB-THREAD builds
87 ;;; 42: (2003-05-22) %NAME slot changed to NAME in
88 ;;; DEFSTRUCT-SLOT-DESCRIPTION
89 ;;; 43: (2003-07-18) Something could easily have changed incompatibly in
90 ;;; recent maintenance, e.g. from (VECTOR NIL)-as-string support.
91 ;;; (And experimental results suggest that compatibility was broken
92 ;;; between about 0.8.1.29 and 0.8.1.39.)
93 ;;; 44: (2003-08-25) various changes leading up to 0.8.3
94 ;;; <dan`b> what happened this month to stalate the fasls?
95 ;;; <Krystof_> I think I renumbered everything again
96 ;;; <Krystof_> simple-array-unsigned-byte-7, probably
97 ;;; <Krystof_> (thanks to pfdietz)
98 ;;; 45: (2003-10-02) I (WHN) incremented the version for the 0.8.4
99 ;;; release because I couldn't immediately convince myself that
100 ;;; .fasl files could never possibly ever refer to the SB-C
101 ;;; CONTINUATION-related data types which were changed
102 ;;; incompatibly in 0.8.3.62.
103 ;;; 46: (2003-11-11) Tim Daly, Jr. (and Christophe Rhodes) reported
104 ;;; .fasl incompatibility on sbcl-devel 2003-11-09.
105 ;;; 47: (2003-11-30) Static variables were rearranged in 0.8.6.11.
106 ;;; 48: (2004-03-01) Renumbered all the widetags to allow for more
107 ;;; microefficiency in sbcl-0.8.8.10
108 ;;; 49: (2004-05-04) Changed implementation of DEFFOO macros and the
109 ;;; functions they expand to.
110 ;;; 50: (2004-05-20) Changed %COMPILER-DEFUN signature again.
112 ;;; the conventional file extension for our fasl files
115 \f
116 ;;;; information about below-Lisp-level linkage
118 ;;; Note:
119 ;;; Assembler routines are named by full Lisp symbols: they
120 ;;; have packages and that sort of native Lisp stuff associated
121 ;;; with them. We can compare them with EQ.
122 ;;; Foreign symbols are named by Lisp STRINGs: the Lisp package
123 ;;; system doesn't extend out to symbols in languages like C.
124 ;;; We want to use EQUAL to compare them.
125 ;;; *STATIC-FOREIGN-SYMBOLS* are static as opposed to "dynamic" (not
126 ;;; as opposed to C's "extern"). The table contains symbols known at
127 ;;; the time that the program was built, but not symbols defined
128 ;;; in object files which have been loaded dynamically since then.
132 \f
133 ;;;; the FOP database
137 ;;; a vector indexed by a FaslOP that yields the FOP's name
140 ;;; a vector indexed by a FaslOP that yields a function of 0 arguments
141 ;;; which will perform the operation
146 \f
147 ;;;; other miscellaneous loading-related stuff
149 \f
150 ;;;; variables
153 #!+sb-doc
157 ;;; the FASL file we're reading from
162 #!+sb-doc
165 ;; Note that CMU CL's default for this was T, and ANSI says it's
166 ;; implementation-dependent. We choose NIL on the theory that it's
167 ;; a nicer default behavior for Unix programs.
168 #!+sb-doc