🚀 go-pugleaf

(This is my third attempt at posting this article with the first
 two were trached by news so hopefully there are no duplicates.)

In working with the GNU C preprocessor, I have come across several
behaviors which appear to deviate from the ANSI C spec (X3.159-1989).
I am hoping to have someone knowledgeable in the spec will confirm or
deny my suspicions. 

1) In chapter 3.8 which discusses C preprocessing, no mention is
   made of what should occur when a '#' directive is encountered
   which does not fall under the preprocessor syntax.  I would
   assume that unrecognized '#' directives would be an error
   in the preprocessing stage.  For example, if the input file
   has '#foo', what is the correct ANSI C behavior?

   The GNU C preprocessor silently passes unrecognized directives to
   the compiler and expects the compiler to declare the error.

2) For line control (3.8.4), the preprocessor must use '#line'.
   An alternate syntax '# digit-sequence' is not legal under ANSI C.

   The GNU C preprocessor will generate '#line' style if compiled
   with OUTPUT_LINE_COMMANDS otherwise it generates the alternate
   syntax.  To be closer to ANSI C compliance, the preprocessor must
   be built with this compile time option defined.  Correct?

3) The ANSI C spec in 3.8.4 states that '#line' directives must be
   of the format:
	# line digit-sequence new-line
   or
	# line digit-sequence "s-char-sequence" new-line

   In 3.8.4 around line 20, the spec does state that the form:
	# line pp-tokens new-line
   is allowed, but after processing the tokens, the syntax must
   fall into one of the first two forms.

   The GNU C preprocessor will generate '#line' directives in the
   format which adds a digit "1" or "2" if entering or exiting an
   included file.  For example:
	# line 34 "myfile.h" 1

   This seems in violation of the ANSI C spec.
-- 
Quentin Barnes
qbarnes@urbana.mcd.mot.com | ..!uiucuxc!udc!qbarnes