CEDET

The make.by file was the only file to ever change the old backquote
style.  I'll guess that the backquote style was lost in the transition
to .by, so everything backquoted will need a minor upgrade.

Eric

>>> David PONCE <david.ponce@...> seems to think that:
>Klaus,
>
>Thanks again for your bug report!
>
>> i a Makefile contains:
>> 
>> include path/to/another/Makefile
>> 
>> then i get the following error:
>> 
>> Wrong type argument: (intergerp, "path/to/another/Makefile")
>
>I tried a simple test case:
>
>-------
>include path/to/another/Makefile path/Test
>
>all: doit test test2
>
>doit:
>
>-------
>
>It didn't failed with the "Wrong type argument" error, but produced a
>bad parse tree.
>
>The following patch seemed to fix that.
>
>Eric, does it look good for you?
>
>If so, I will check it in.
>
>David
>
>Index: make.by
>===================================================================
>RCS file: /cvsroot/cedet/cedet/semantic/bovine/make.by,v
>retrieving revision 1.8
>diff -c -r1.8 make.by
>*** make.by	2 Aug 2003 08:13:14 -0000	1.8
>--- make.by	5 Nov 2003 15:27:13 -0000
>***************
>*** 62,72 ****
>  	 ;
>  
>  variable: symbol opt-whitespace equals opt-whitespace element-list
>! 	  (VARIABLE-TAG ,$1 nil ,$5)
>  	;
>  
>  rule: targets opt-whitespace colons opt-whitespace element-list commands
>!       (FUNCTION-TAG ,$1 nil ,$5)
>      ;
>  
>  targets: target opt-whitespace targets
>--- 62,72 ----
>  	 ;
>  
>  variable: symbol opt-whitespace equals opt-whitespace element-list
>! 	  (VARIABLE-TAG ,$1 nil (car ,$5))
>  	;
>  
>  rule: targets opt-whitespace colons opt-whitespace element-list commands
>!       (FUNCTION-TAG ,$1 nil (car ,$5))
>      ;
>  
>  targets: target opt-whitespace targets
>***************
>*** 106,112 ****
>  	   ;
>  
>  include: INCLUDE whitespace element-list
>! 	 (INCLUDE-TAG ,$3 nil)
>         ;
>  
>  equals: COLON EQUAL ()
>--- 106,112 ----
>  	   ;
>  
>  include: INCLUDE whitespace element-list
>! 	 (INCLUDE-TAG (car ,$3) nil)
>         ;
>  
>  equals: COLON EQUAL ()
>***************
>*** 119,125 ****
>        ;
>  
>  element-list: elements newline
>! 	      ( ,@$1 )
>  	    ;
>  
>  elements: element whitespace elements
>--- 119,125 ----
>        ;
>  
>  element-list: elements newline
>! 	      ( (apply 'append ,$1) )
>  	    ;
>  
>  elements: element whitespace elements
>
>

-- 
          Eric Ludlam:                 zappo@..., eric@...
   Home: http://www.ludlam.net            Siege: http://www.siege-engine.com
Emacs: http://cedet.sourceforge.net               GNU: http://www.gnu.org

Hi Eric,

> The make.by file was the only file to ever change the old backquote
> style.  I'll guess that the backquote style was lost in the
> transition to .by, so everything backquoted will need a minor
> upgrade.

No, I checked that the backquote style works as it previously did with
BNF files.

In fact what has changed in the new version, is the tag expander.
Previously `semantic-expand-make-nonterminal' considered two
different cases of expansion:

- function tags of the form:

    (("function-name") 'function ....)

- include tags of the form

   ((("include-name1") ("include-name2") ...) 'include ...)

The new expander `semantic-make-expand-tag' is probably
over-simplified and considers that function and include tags have the
same form (the first form above).

So there is two possibilities to fix tag expansion:

- The first one is to change the grammar to return include tags
  having the same form as function tags:

  (("include-name1" "include-name2" ...) 'include ...)

  This is what I did in my previous patch to make.by.

  The inconvenient is that it changed the value returned for
  `element-list' nonterminal which is also part of variable and
  function tags.

- The second possibility is to fix `semantic-make-expand-tag' and
  left the grammar untouched.

From a beta release perspective, where compatibility is essential, I
think the second solution is probably the best one.

Following is a new patch to fix `semantic-make-expand-tag'.

If you don't have objection I will check it in the cedet-1p0beta1
branch.

David

2003-11-06  David Ponce  <david@...>

	* semantic/bovine/semantic-make.el

	(semantic-make-expand-tag): Fix expansion of include tags.

Index: semantic/bovine/semantic-make.el
===================================================================
RCS file: /cvsroot/cedet/cedet/semantic/bovine/semantic-make.el,v
retrieving revision 1.10
diff -c -r1.10 semantic-make.el
*** semantic/bovine/semantic-make.el	2 Aug 2003 08:14:24 -0000	1.10
--- semantic/bovine/semantic-make.el	6 Nov 2003 09:07:42 -0000
***************
*** 64,76 ****
  
  (defun semantic-make-expand-tag (tag)
    "Expand TAG into a list of equivalent tags, or nil."
!   (let (name xpand)
!     (when (and (memq (semantic-tag-class tag) '(function include))
!                (listp (setq name (semantic-tag-name tag))))
!       (while name
!         (setq xpand (cons (semantic-tag-clone tag (car name)) xpand)
!               name  (cdr name)))
!       xpand)))
  
  ;;;###autoload
  (defun semantic-default-make-setup ()
--- 64,84 ----
  
  (defun semantic-make-expand-tag (tag)
    "Expand TAG into a list of equivalent tags, or nil."
!   (let ((name (semantic-tag-name tag))
!         xpand)
!     (when (consp name)
!       (cond
!        ((semantic-tag-of-class-p tag 'function)
!         (while name
!           (setq xpand (cons (semantic-tag-clone tag (car name)) xpand)
!                 name  (cdr name)))
!         )
!        ((semantic-tag-of-class-p tag 'include)
!         (while name
!           (setq xpand (cons (semantic-tag-clone tag (caar name)) xpand)
!                 name  (cdr name)))
!         )))
!     xpand))
  
  ;;;###autoload
  (defun semantic-default-make-setup ()

>>> David PONCE <david.ponce@...> seems to think that:
  [ ... ]
>Following is a new patch to fix `semantic-make-expand-tag'.
>
>If you don't have objection I will check it in the cedet-1p0beta1
>branch.
>
>David
>
>2003-11-06  David Ponce  <david@...>
>
>	* semantic/bovine/semantic-make.el
>
>	(semantic-make-expand-tag): Fix expansion of include tags.
  [ ... ]

That looks good to me.  Thanks.

Eric

-- 
          Eric Ludlam:                 zappo@..., eric@...
   Home: http://www.ludlam.net            Siege: http://www.siege-engine.com
Emacs: http://cedet.sourceforge.net               GNU: http://www.gnu.org