1 ;;;; the bare essentials of compiler error handling
3 ;;;; (Logically, this might belong in early-c.lisp, since it's stuff
4 ;;;; which might as well be visible to all compiler code. However,
5 ;;;; physically its DEFINE-CONDITION forms depend on the condition
6 ;;;; system being set up before it can be cold loaded, so we keep it
7 ;;;; in this separate, loaded-later file instead of in early-c.lisp.)
9 ;;;; This software is part of the SBCL system. See the README file for
10 ;;;; more information.
12 ;;;; This software is derived from the CMU CL system, which was
13 ;;;; written at Carnegie Mellon University and released into the
14 ;;;; public domain. The software is in the public domain and is
15 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
16 ;;;; files for more information.
20 ;;;; error-handling definitions which are easy to define early and
21 ;;;; which are nice to have visible everywhere
23 ;;; a function that is called to unwind out of COMPILER-ERROR
24 (declaim (type (function (&optional condition
) nil
) *compiler-error-bailout
*))
25 (defvar *compiler-error-bailout
*)
27 ;;; an application programmer's error caught by the compiler
29 ;;; We want a separate condition for application programmer errors so
30 ;;; that we can distinguish them from system programming errors (bugs
31 ;;; in SBCL itself). Application programmer errors should be caught
32 ;;; and turned into diagnostic output and a FAILURE-P return value
33 ;;; from COMPILE or COMPILE-FILE. Bugs in SBCL itself throw us into
36 ;;; A further word or two of explanation might be warranted here,
37 ;;; since I (CSR) have spent the last day or so wandering in a
38 ;;; confused daze trying to get this to behave nicely before finally
39 ;;; hitting on the right solution.
41 ;;; These objects obey a slightly involved protocol in order to
42 ;;; achieve the right dynamic behaviour. If we signal a
43 ;;; COMPILER-ERROR from within the compiler, we want that the
44 ;;; outermost call to COMPILE/COMPILE-FILE cease attempting to compile
45 ;;; the code in question and instead compile a call to signal a
46 ;;; PROGRAM-ERROR. This is achieved by resignalling the condition
47 ;;; from within the handler, so that the condition travels up the
48 ;;; handler stack until it finds the outermost handler. Why the
49 ;;; outermost? Well, COMPILE-FILE could call EVAL from an EVAL-WHEN,
50 ;;; which could recursively call COMPILE, which could then signal an
51 ;;; error; we want the inner EVAL not to fail so that we can go on
52 ;;; compiling, so it's the outer COMPILE-FILE that needs to replace
53 ;;; the erroneous call with a call to ERROR.
55 ;;; This resignalling up the stack means that COMPILER-ERROR should
56 ;;; not be a generalized instance of ERROR, as otherwise code such as
57 ;;; (IGNORE-ERRORS (DEFGENERIC IF (X))) will catch and claim to handle
58 ;;; the COMPILER-ERROR. So we make COMPILER-ERROR inherit from
59 ;;; CONDITION instead, as of sbcl-0.8alpha.0.2x, so that unless
60 ;;; the user claims to be able to handle general CONDITIONs (and if he
61 ;;; does, he deserves what's going to happen :-) [ Note: we don't make
62 ;;; COMPILER-ERROR inherit from SERIOUS-CONDITION, because
63 ;;; conventionally SERIOUS-CONDITIONs, if unhandled, end up in the
64 ;;; debugger; although the COMPILER-ERROR might well trigger an entry
65 ;;; into the debugger, it won't be the COMPILER-ERROR itself that is
66 ;;; the direct cause. ]
68 ;;; So, what if we're not inside the compiler, then? Well, in that
69 ;;; case we're in the evaluator, so we want to convert the
70 ;;; COMPILER-ERROR into a PROGRAM-ERROR and signal it immediately. We
71 ;;; have to signal the PROGRAM-ERROR from the dynamic environment of
72 ;;; attempting to evaluate the erroneous code, and not from any
73 ;;; exterior handler, so that user handlers for PROGRAM-ERROR and
74 ;;; ERROR stand a chance of running, in e.g. (IGNORE-ERRORS
75 ;;; (DEFGENERIC IF (X))). So this is where the SIGNAL-PROGRAM-ERROR
76 ;;; restart comes in; the handler in EVAL-IN-LEXENV chooses this
77 ;;; restart if it believes that the compiler is not present (which it
78 ;;; tests using the BOUNDPness of *COMPILER-ERROR-BAILOUT*). The
79 ;;; restart executes in the dynamic environment of the original
80 ;;; COMPILER-ERROR call, and all is well.
83 (define-condition compiler-error
(encapsulated-condition) ()
84 (:report
(lambda (condition stream
)
85 (print-object (encapsulated-condition condition
) stream
))))
87 ;;; Signal the appropriate condition. COMPILER-ERROR calls the bailout
88 ;;; function so that it never returns (but compilation continues).
89 (declaim (ftype (function (t &rest t
) nil
) compiler-error
))
90 (defun compiler-error (datum &rest arguments
)
91 (let ((condition (coerce-to-condition datum arguments
92 'simple-program-error
'compiler-error
)))
95 (cerror "Replace form with call to ERROR."
98 (funcall *compiler-error-bailout
* condition
)
99 (bug "Control returned from *COMPILER-ERROR-BAILOUT*."))
101 (error condition
)))))
103 (defun compiler-warn (datum &rest arguments
)
104 (apply #'warn datum arguments
)
107 (defun compiler-style-warn (datum &rest arguments
)
108 (apply #'style-warn datum arguments
)
111 (defun make-compiler-error-form (condition source
)
112 `(error 'compiled-program-error
113 :message
,(princ-to-string condition
)
114 :source
,(princ-to-string source
)))
116 ;;; Fatal compiler errors. We export FATAL-COMPILER-ERROR as an
117 ;;; interface for errors that kill the compiler dead
119 ;;; These are not a COMPILER-ERRORs, since we don't try to recover
120 ;;; from them and keep chugging along, but instead immediately bail
121 ;;; out of the entire COMPILE-FILE.
123 (define-condition fatal-compiler-error
(encapsulated-condition)
126 ;;; the condition of COMPILE-FILE being unable to READ from the
129 ;;; (The old CMU CL code did try to recover from this condition, but
130 ;;; the code for doing that was messy and didn't always work right.
131 ;;; Since in Common Lisp the simple act of reading and compiling code
132 ;;; (even without ever loading the compiled result) can have side
133 ;;; effects, it's a little scary to go on reading code when you're
134 ;;; deeply confused, so we violate what'd otherwise be good compiler
135 ;;; practice by not trying to recover from this error and bailing out
137 (define-condition input-error-in-compile-file
(fatal-compiler-error)
138 (;; the position where the bad READ began, or NIL if unavailable,
139 ;; redundant, or irrelevant
140 (position :reader input-error-in-compile-file-position
144 (lambda (condition stream
)
146 "~@<~S failure in ~S~@[ at character ~W~]: ~2I~_~A~:>"
149 (input-error-in-compile-file-position condition
)
150 (encapsulated-condition condition
)))))