[Sbcl-commits] CVS: sbcl/tests reader.pure.lisp,1.2,1.3

[William H. N. <wn...@us...>]

Update of /cvsroot/sbcl/sbcl/tests
In directory usw-pr-cvs1:/tmp/cvs-serv21671/tests

Modified Files:
	reader.pure.lisp 
Log Message:
0.7.3.16:
	(These changes might leave SET-SYNTAX-FROM-CHAR messed up. I'll
		try to fix that soon, or else add some tests to verify
		that it isn't too messed up.)
	ANSIfying FROB-MACRO-CHARACTER functions...
	...made GET-MACRO-CHARACTER return NIL for non-macro characters
	...made SET-MACRO-CHARACTER handle NIL values
	...used %COERCE-CALLABLE-TO-FUN instead of COERCE ... 'FUNCTION
	Some of the "Unconstituentize some of them" operations in 
		!COLD-INIT-STANDARD-READTABLE seem to've become no-ops,
		so delete them.
	It seems strange that ASCII RUBOUT should be treated as
		whitespace, so that e.g. "(+ 12)" EVALs to 3. So
		I deleted the customization of its entry in
		*STANDARD-READTABLE*, and now it's treated as just
		another character, which seems less surprising.


Index: reader.pure.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/tests/reader.pure.lisp,v
retrieving revision 1.2
retrieving revision 1.3
diff -C2 -d -r1.2 -r1.3
*** reader.pure.lisp	12 Apr 2002 12:15:56 -0000	1.2
--- reader.pure.lisp	14 May 2002 04:27:35 -0000	1.3
***************
*** 17,28 ****
  
  ;;; Prior to sbcl-0.7.2.10, SBCL disobeyed the ANSI requirements on
! ;;; returning NIL for unset dispatch-macro-character functions (bug
  ;;; 151, fixed by Alexey Dejenka sbcl-devel "bug 151" 2002-04-12)
  (assert (not (get-dispatch-macro-character #\# #\{)))
  (assert (not (get-dispatch-macro-character #\# #\0)))
! ;;; and we might as well test that we don't have any cross-compilation
  ;;; shebang residues left...
  (assert (not (get-dispatch-macro-character #\# #\!)))
! ;;; also test that all the illegal sharp macro characters are
  ;;; recognized as being illegal.
  (loop for char in '(#\Backspace #\Tab #\Newline #\Linefeed
--- 17,28 ----
  
  ;;; Prior to sbcl-0.7.2.10, SBCL disobeyed the ANSI requirements on
! ;;; returning NIL for unset dispatch-macro-character functions. (bug
  ;;; 151, fixed by Alexey Dejenka sbcl-devel "bug 151" 2002-04-12)
  (assert (not (get-dispatch-macro-character #\# #\{)))
  (assert (not (get-dispatch-macro-character #\# #\0)))
! ;;; And we might as well test that we don't have any cross-compilation
  ;;; shebang residues left...
  (assert (not (get-dispatch-macro-character #\# #\!)))
! ;;; Also test that all the illegal sharp macro characters are
  ;;; recognized as being illegal.
  (loop for char in '(#\Backspace #\Tab #\Newline #\Linefeed
***************
*** 32,33 ****
--- 32,53 ----
  (assert (not (ignore-errors (get-dispatch-macro-character #\! #\0)
                              t)))
+ 
+ ;;; In sbcl-0.7.3, GET-MACRO-CHARACTER and SET-MACRO-CHARACTER didn't
+ ;;; use NIL to represent the no-macro-attached-to-this-character case
+ ;;; as ANSI says they should. (This problem is parallel to the
+ ;;; GET-DISPATCH-MACRO misbehavior fixed in sbcl-0.7.2.10, but
+ ;;; was fixed a little later.)
+ (dolist (customizable-char
+ 	 ;; According to ANSI "2.1.4 Character Syntax Types", these
+ 	 ;; characters are reserved for the programmer.
+ 	 '(#\? #\! #\[ #\] #\{ #\}))
+   ;; So they should have no macro-characterness.
+   (multiple-value-bind (macro-fun non-terminating-p)
+       (get-macro-character customizable-char)
+     (assert (null macro-fun))
+     ;; Also, in a bit of ANSI weirdness, NON-TERMINATING-P can be
+     ;; true only when MACRO-FUN is true. (When the character
+     ;; is not a macro character, it can be embedded in a token,
+     ;; so it'd be more logical for NON-TERMINATING-P to be T in
+     ;; this case; but ANSI says it's NIL in this case.
+     (assert (null non-terminating-p))))