CEDET

Hi Eric,

I recently had the opportunity to work a little bit more with XEmacs
than I usually do, and discovered that semantic regularly failed
because of the missing `compare-strings' function.

A quick grep across CEDET packages shows that `compare-strings' is
used in these libraries:

./semantic/semantic-complete.el:421:
./semantic/semantic-find.el:290:
./semantic/semantic-sort.el:54:

An Elisp implementation of `compare-strings' is in the eieio directory
but no more used there.

I propose to move that `compare-strings' implementation in a new
cedet-compat.el library in the common directory, and make
`compare-strings' an auto loaded function, so it will be automatically
provided when necessary.  In the future we could move other
compatibility stuff (like overlay/extent) to the cedet-compat library.

Following is a change log, the source of cedet-compat.el, and a patch
to include that file in the "common" project.

What do you think?

David

2003-09-30  David Ponce  <david@...>

	* common/cedet-compat.el

	New file.

	* common/Makefile

	Re-generate.

	* common/Project.ede

	(common): Add cedet-compat.el to target sources.

	* eieio/compare-strings.el

	Remove.

;;; cedet-compat.el --- Compatibility across (X)Emacs versions

;; Copyright (C) 2003 David Ponce

;; Author: David Ponce <david@...>
;; Maintainer: David Ponce <david@...>
;; Keywords: compatibility
;; X-RCS: $Id$

;; This file is not part of Emacs

;; This program is free software; you can redistribute it and/or
;; modify it under the terms of the GNU General Public License as
;; published by the Free Software Foundation; either version 2, or (at
;; your option) any later version.

;; This program is distributed in the hope that it will be useful, but
;; WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
;; General Public License for more details.

;; You should have received a copy of the GNU General Public License
;; along with this program; see the file COPYING.  If not, write to
;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
;; Boston, MA 02111-1307, USA.

;;; Commentary:
;;
;; This library provides functions to allow running CEDET packages on
;; a variety of [X]Emacs versions.

;;; History:
;;

;;; Code:

;; XEmacs does not have the `compare-strings' function.  Here is an
;; implementation in Emacs Lisp, derived from the C implementation
;; found in src/fns.c, in GNU Emacs 21.3.1 sources.
;;;###autoload
(defun compare-strings (str1 start1 end1 str2 start2 end2 &optional ignore-case)
   "Compare the contents of two strings.
In string STR1, skip the first START1 characters and stop at END1.
In string STR2, skip the first START2 characters and stop at END2.
END1 and END2 default to the full lengths of the respective strings.

Case is significant in this comparison if IGNORE-CASE is nil.

The value is t if the strings (or specified portions) match.
If string STR1 is less, the value is a negative number N;
   - 1 - N is the number of characters that match at the beginning.
If string STR1 is greater, the value is a positive number N;
   N - 1 is the number of characters that match at the beginning."
   (or start1 (setq start1 0))
   (or start2 (setq start2 0))
   (setq end1 (if end1
                  (min end1 (length str1))
                (length str1)))
   (setq end2 (if end2
                  (min end2 (length str2))
                (length str2)))
   (let ((i1 start1)
         (i2 start2)
         result c1 c2)
     (while (and (not result) (< i1 end1) (< i2 end2))
       (setq c1 (aref str1 i1)
             c2 (aref str2 i2)
             i1 (1+ i1)
             i2 (1+ i2))
       (if ignore-case
           (setq c1 (upcase c1)
                 c2 (upcase c2)))
       (setq result (cond ((< c1 c2) (- i1))
                          ((> c1 c2) i1))))
     (or result
         (cond ((< i1 end1) (1+ (- i1 start1)))
               ((< i2 end2) (1- (- start1 i1)))
               (t)))
     ))

(provide 'cedet-compat)

;;; cedet-compat.el ends here
Index: common/Makefile
===================================================================
RCS file: /cvsroot/cedet/cedet/common/Makefile,v
retrieving revision 1.2
diff -c -r1.2 Makefile
*** common/Makefile	24 Sep 2003 08:04:30 -0000	1.2
--- common/Makefile	30 Sep 2003 20:18:10 -0000
***************
*** 16,22 ****
   init_LISP=cedet-load.el
   EMACS=emacs
   setup_LISP=cedet.el
! common_LISP=cedet-autogen.el ezimage.el inversion.el pprint.el sformat.el working.el
   VERSION=1.0beta1
   DISTDIR=$(top)common-$(VERSION)

--- 16,22 ----
   init_LISP=cedet-load.el
   EMACS=emacs
   setup_LISP=cedet.el
! common_LISP=cedet-autogen.el cedet-compat.el ezimage.el inversion.el pprint.el sformat.el working.el
   VERSION=1.0beta1
   DISTDIR=$(top)common-$(VERSION)

Index: common/Project.ede
===================================================================
RCS file: /cvsroot/cedet/cedet/common/Project.ede,v
retrieving revision 1.2
diff -c -r1.2 Project.ede
*** common/Project.ede	24 Sep 2003 08:04:09 -0000	1.2
--- common/Project.ede	30 Sep 2003 20:18:10 -0000
***************
*** 24,30 ****
      (ede-proj-target-elisp "common"
       :name "common"
       :path ""
!     :source '("cedet-autogen.el" "ezimage.el" "inversion.el" "pprint.el" "sformat.el" "working.el")
       )
      )
     :web-site-url "http://cedet.sourceforge.net/"
--- 24,30 ----
      (ede-proj-target-elisp "common"
       :name "common"
       :path ""
!     :source '("cedet-autogen.el" "cedet-compat.el" "ezimage.el" "inversion.el" "pprint.el" "sformat.el" "working.el")
       )
      )
     :web-site-url "http://cedet.sourceforge.net/"

Hi all,

In semantic 2.x/XEmacs, there is a timer that starts up after about 4
seconds of idle time, as I understand it, to start checking for buffer
cahnges and reparse, etc.

Once it kicks in, the mouse starts changing from the text icon to the
clock icon and back almost twice a second. I'm having a hard time
getting a trace of where the cursor change might be occurring, as any
user event (mouse, command, key) causes the activity to cease for
another 4 seconds.

The constant flickering is very distracting when pausing to consider
your next programming feat. I'm assuming that in Emacs this does not
occur? 

I'm running debian linux. 

Thanks,
Stuart Popejoy

I updated EIEIO and COGRE to have beta1 versions to match the
beta-ness of CEDET.

I noticed that the common directory has no Project.ede, nor a Makefile
in CVS.  I happen to have one that I've apparently been maintaining
and forgot to check in.  Unless someone knows of something I am
missing, I will register these in CVS.

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

Hi Eric,

>>>Perhaps the `common' directory could just be called `cedet' and mark
>>>it as the location for miscellaneous bits not large enough to warrant
>>>a project directory, but still quite specific to CEDET.
>>
>>[...]
>>
>>I agree with that (I thought to it too).  An advantage is that the
>>library cedet.el already exists, and provide the cedet-version
>>info. needed by inversion (IMO it makes sense that the cedet version
>>serve as reference for common stuff).
>>
>>Is there a simple way to do that renaming in CVS?  Or should we create
>>the new cedet directory then add the content of the common directory
>>into it, and finally remove the common directory from the repository.
> 
>   [ ... ]
> 
> The only way to do this CVS change in one pass is if you have access
> to the CVS repository file system.  (ie, we could download a
> repository backup, make the change, and request an upload via the
> sourceforge system administrators...  Ick!)
> 
> A description of doing it via CVS commands is in the CVS info pages
> and is pretty easy too.  If you intend to do this, that would be
> great.

I finally found a better solution which doesn't require to rename the
common subdirectory.  The name "common" is really better than "cedet"
which can confuse people, because cedet is already used for the main
project directory.

In fact it was relatively easy.  I just modified a little
`inversion-add-to-load-path' to actually use the INSTALLDIR optional
argument, so that it is now possible to have a package in a
subdirectory with a different name.

I also changed cedet.el to take into account that.

In conclusion, common stuff are handled in a "pseudo" cedet package
hosted in the "common" subdirectory.

Of course, there is no more common*.el files, they are now cedet.el,
cedet-load.el and the generated cedet-loaddefs.el.

You will find the new files in the attached tarball.  Here is the
patch for existing files.

Thoughts?

Index: Makefile
===================================================================
RCS file: /cvsroot/cedet/cedet/Makefile,v
retrieving revision 1.2
diff -c -r1.2 Makefile
*** Makefile	17 Sep 2003 08:54:34 -0000	1.2
--- Makefile	23 Sep 2003 10:48:10 -0000
***************
*** 29,34 ****
--- 29,35 ----
  
  ## The CEDET's packages installed
  CEDET_PACKAGES=\
+ common \
  ede \
  speedbar \
  eieio \
***************
*** 46,54 ****
  #RM = rm -f
  
  ############### Internal part of the Makefile ###############
- __LOADDEFS=$(patsubst %,%-loaddefs.el,$(CEDET_PACKAGES))
  __BUILD_AUTOLOADS=$(patsubst %,%-autoloads,$(CEDET_PACKAGES))
! __CLEAN_AUTOLOADS=$(patsubst %,clean-%,$(__LOADDEFS))
  __DOMAKE=$(MAKE) $(MFLAGS) EMACS="$(EMACS)" SHELL="$(SHELL)"
  
  ## Build
--- 47,54 ----
  #RM = rm -f
  
  ############### Internal part of the Makefile ###############
  __BUILD_AUTOLOADS=$(patsubst %,%-autoloads,$(CEDET_PACKAGES))
! __CLEAN_AUTOLOADS=$(patsubst %,clean-%,$(__BUILD_AUTOLOADS))
  __DOMAKE=$(MAKE) $(MFLAGS) EMACS="$(EMACS)" SHELL="$(SHELL)"
  
  ## Build
***************
*** 82,88 ****
  
  .PHONY: $(__CLEAN_AUTOLOADS)
  $(__CLEAN_AUTOLOADS):
! 	$(RM) $(CEDET_HOME)/$(word 2,$(subst -, ,$@))/$(subst clean-,,$@)
  
  .PHONY: clean-grammars
  clean-grammars:
--- 82,90 ----
  
  .PHONY: $(__CLEAN_AUTOLOADS)
  $(__CLEAN_AUTOLOADS):
! 	$(FIND) $(CEDET_HOME)/$(word 2,$(subst -, ,$@)) -type f \
! 	-name "*-loaddefs.el" \
! 	-print -exec $(RM) {} \;
  
  .PHONY: clean-grammars
  clean-grammars:
Index: common/cedet.el
===================================================================
RCS file: /cvsroot/cedet/cedet/common/cedet.el,v
retrieving revision 1.3
diff -c -r1.3 cedet.el
*** common/cedet.el	17 Sep 2003 08:56:12 -0000	1.3
--- common/cedet.el	23 Sep 2003 10:48:11 -0000
***************
*** 76,92 ****
  (eval-when-compile
    (require 'cl))
  
! (defconst cedet-version "1.0"
    "Current version of CEDET.")
  
  (defconst cedet-packages
!   '(
!     ;;PACKAGE   MIN-VERSION
!     (cogre      "0.4"       )
!     (ede        "1.0beta3"  )
!     (eieio      "0.18"      )
!     (semantic   "2.0beta1"  )
!     (speedbar   "0.15beta1" )
      )
    "Table of CEDET packages to install.")
  
--- 76,93 ----
  (eval-when-compile
    (require 'cl))
  
! (defconst cedet-version "1.0beta1"
    "Current version of CEDET.")
  
  (defconst cedet-packages
!   `(
!     ;;PACKAGE   MIN-VERSION      INSTALLDIR
!     (cedet      ,cedet-version   "common"  )
!     (cogre      "0.4"                      )
!     (ede        "1.0beta3"                 )
!     (eieio      "0.18"                     )
!     (semantic   "2.0beta1"                 )
!     (speedbar   "0.15beta1"                )
      )
    "Table of CEDET packages to install.")
  
***************
*** 102,115 ****
    (require 'inversion)
    
    ;; Go up to the parent "<INSTALL-DIR>/cedet" directory.
!   (let ((default-directory (expand-file-name "..")))
  
      ;; Add the CEDET packages subdirectories to the `load-path' if
      ;; necessary, and do specific setup.
!     (dolist (package cedet-packages)
!       (inversion-add-to-load-path (car package) (cadr package))
        (condition-case err
!           (require (intern (format "%s-load" (car package))))
          (error
           (message "%s" (error-message-string err)))))
      ))
--- 103,122 ----
    (require 'inversion)
    
    ;; Go up to the parent "<INSTALL-DIR>/cedet" directory.
!   (let ((default-directory (expand-file-name ".."))
!         package min-version installdir)
  
      ;; Add the CEDET packages subdirectories to the `load-path' if
      ;; necessary, and do specific setup.
!     (dolist (package-spec cedet-packages)
!       (setq package     (nth 0 package-spec)
!             min-version (nth 1 package-spec)
!             installdir  (nth 2 package-spec))
!       (when installdir
!         (setq installdir (expand-file-name installdir)))
!       (inversion-add-to-load-path package min-version installdir)
        (condition-case err
!           (require (intern (format "%s-load" package)))
          (error
           (message "%s" (error-message-string err)))))
      ))
Index: common/inversion.el
===================================================================
RCS file: /cvsroot/cedet/cedet/common/inversion.el,v
retrieving revision 1.19
diff -c -r1.19 inversion.el
*** common/inversion.el	6 Sep 2003 19:41:41 -0000	1.19
--- common/inversion.el	23 Sep 2003 10:48:11 -0000
***************
*** 359,386 ****
    "Add the PACKAGE path to `load-path' if necessary.
  MINIMUM is the minimum version requirement of PACKAGE.
  Optional argument INSTALLDIR is the base directory where PACKAGE is
! installed.  It defaults to `default-directory'.
! SUBDIRS are PACKAGE sub-directories to add to `load-path', following
! the main INSTALLDIR/PACKAGE path."
    (let ((ver (inversion-find-version package)))
      ;; If PACKAGE not found or a bad version already in `load-path',
      ;; prepend the new PACKAGE path, so it will be loaded first.
      (when (or (not ver)
  	      (inversion-check-version (car ver) (cdr ver) minimum))
!       (let* ((default-directory (or installdir default-directory))
! 	     (path (expand-file-name (format "./%s" package)))
! 	     subpath)
! 	(when (file-directory-p path)
  	  ;; Add SUBDIRS
  	  (while subdirs
! 	    (setq subpath (format "%s/%s" path (car subdirs))
  		  subdirs (cdr subdirs))
! 	    (when (file-directory-p subpath)
! 	      (message "%S added to `load-path'" subpath)
! 	      (add-to-list 'load-path subpath)))
  	  ;; Add the main path
! 	  (message "%S added to `load-path'" path)
! 	  (add-to-list 'load-path path))))))
  
  ;;; Inversion tests
  ;;
--- 359,387 ----
    "Add the PACKAGE path to `load-path' if necessary.
  MINIMUM is the minimum version requirement of PACKAGE.
  Optional argument INSTALLDIR is the base directory where PACKAGE is
! installed.  It defaults to `default-directory'/PACKAGE.
! SUBDIRS are sub-directories to add to `load-path', following the main
! INSTALLDIR path."
    (let ((ver (inversion-find-version package)))
      ;; If PACKAGE not found or a bad version already in `load-path',
      ;; prepend the new PACKAGE path, so it will be loaded first.
      (when (or (not ver)
  	      (inversion-check-version (car ver) (cdr ver) minimum))
!       (let* ((default-directory
! 	       (or installdir
! 		   (expand-file-name (format "./%s" package))))
! 	     subdir)
! 	(when (file-directory-p default-directory)
  	  ;; Add SUBDIRS
  	  (while subdirs
! 	    (setq subdir  (expand-file-name (car subdirs))
  		  subdirs (cdr subdirs))
! 	    (when (file-directory-p subdir)
! 	      (message "%S added to `load-path'" subdir)
! 	      (add-to-list 'load-path subdir)))
  	  ;; Add the main path
! 	  (message "%S added to `load-path'" default-directory)
! 	  (add-to-list 'load-path default-directory))))))
  
  ;;; Inversion tests
  ;;

Hi Eric,

For now CEDET lacks a true CEDET's common package.  Here is a proposal
to supply one.

First I created these new files (in the attached tarball):

common.el
  This file just provide the `common-version' information for
  inversion.

common-load.el
  Initialize common stuff.  For now it just loads common autoloads
  from the generated common-loaddefs.el file.

Project.ede
  The CEDET's common stuff project used to generate a Makefile.

Then I made minor changes to the main Makefile and cedet.el to take
into account the common package (see the patch at end).

Also, it would be good to move sformat.el and working.el from semantic
to the common package which is now mandatory.

What do you think?

Index: Makefile
===================================================================
RCS file: /cvsroot/cedet/cedet/Makefile,v
retrieving revision 1.2
diff -c -r1.2 Makefile
*** Makefile	17 Sep 2003 08:54:34 -0000	1.2
--- Makefile	22 Sep 2003 13:49:28 -0000
***************
*** 29,34 ****
--- 29,35 ----
  
  ## The CEDET's packages installed
  CEDET_PACKAGES=\
+ common \
  ede \
  speedbar \
  eieio \
Index: common/cedet.el
===================================================================
RCS file: /cvsroot/cedet/cedet/common/cedet.el,v
retrieving revision 1.3
diff -c -r1.3 cedet.el
*** common/cedet.el	17 Sep 2003 08:56:12 -0000	1.3
--- common/cedet.el	22 Sep 2003 13:49:28 -0000
***************
*** 82,87 ****
--- 82,88 ----
  (defconst cedet-packages
    '(
      ;;PACKAGE   MIN-VERSION
+     (common     "0.1"       )
      (cogre      "0.4"       )
      (ede        "1.0beta3"  )
      (eieio      "0.18"      )

Hi Klaus,

> i found on the http://www.xref-tech.com/ <http://www.xref-tech.com/>
> the complete ANSI C++-grammar written in YACC-syntax!  Could this be
> used as basis for a semantic-2.0 C++-grammar for the new LALR-wisent
> parser....

Unfortunately not really :-(

ANSI C/C++ grammars are not LALR(1), and, as stated on the
xref-tech.com site, the C++ grammar presented is just:

"A browsable ANSI C++ grammar in YACC format.
 This is not an executable grammar for YACC (due to ambiguities)!
 It is simply the ANSI standard rewritten into YACC syntax."

A such grammar needs a lot of work to become acceptable by YACC, Bison
or Wisent.

I started to work on a LALR C grammar for Wisent (see wisent-c stuff
in CVS).  It is a non-trivial task, and I don't hope to have something
that works for Semantic 2.0.

Of course, any help will be appreciated ;-)

David

P.S.: It seems that even the GCC developers have abandoned the parser
generated from a Bison C++ grammar, because it was too hard to
maintain, and wrote an hand-crafted C++ parser for newer versions of
GCC!

Hi Eric,

[...] 
>>What do you think of adopting a common rule for naming
>>autoloads
>>files?
[...] 
> Sounds good to me.

Done.

> When I am able to compute again, I can make the
> constructor for new autoload targets take the project
> name and add loaddefs to the end.  The lack of this
> feature added the problem.

That would be nice!

>>way semantic is
>>loaded, and provide a <package>-load.el file for each
>>CEDET package.
>>Initially it would just load the package autoloads. 
[...]
> I like that idea too.

Done.

>>Finally, what do you think of removing the following
>>files _that seems
>>no more used_ from the repository:
>>
>>semantic/semantic-autogen.el
>>semantic/bovine/erlang.bnf
>>semantic/wisent/Wisent.mk  (need update of Project.ede)
[...]

Done.

All the CEDET packages are now setup in a consistent way:

(require 'PACKAGE-load) 

  -> (load "PACKAGE-loaddefs")
  -> Do other setup things.

I changed common/cedet.el so that it automatically requires the
PACKAGE-load feature.

All the Project.ede/Makefile are updated.

David

2003-09-17  David Ponce  <david@...>

	* Makefile

	Compute name of autoloads files.  Use explicit list of targets
	instead of pattern rules.

	* common/cedet.el

	(cedet-packages): No more need an explicit setup file to load.
	Automatically require the PACKAGE-load feature.

	* cogre/cogre-load.el

	New file.
	
	* cogre/Makefile

	Re-generated.

	* cogre/Project.ede

	(autoloads): Change cogre-defs.el by cogre-loaddefs.el.

	* ede/ede.el

	(ede-load): Require it.  No more load ede-loaddefs.
	
	* ede/ede-load.el

	New file.
	
	* ede/Makefile

	Re-generated.

	* eieio/eieio-load.el

	New file.
	
	* eieio/Makefile

	Re-generated.

	* eieio/Project.ede
	
	(autoloads): Change eieio-defs.el by eieio-loaddefs.el.
	
	* semantic/semantic-fw.el

	Load semantic-loaddefs instead of semantic-al.
	
	* semantic/Makefile

	Re-generated.

	* semantic/Project.ede
	
	(autoloads): Change semantic-al.el by semantic-loaddefs.el.
	
	* speedbar/speedbar-load.el

	New file.
	
	* speedbar/Project.ede
	
	(autoloads): Change speedbar-defs.el by speedbar-loaddefs.el.
	
	* speedbar/Makefile

	Re-generated.

Looks good to me.  Thanks

Eric

On Tue, 16 Sep 2003 20:25:29 +0200, David Ponce wrote:

> 
> > My computer is dead.  Would someone mind applying
this
> > patch for me?  (Unless, of course, you can think of
a
> > better way.) Thanks.
> 
> What a pity :-(
> 
> Do you agree with this version:
> 
> (defun semantic-changes-in-region (start end &optional
> buffer)
>    "Find change overlays which exist in whole or in
> part between START and END.
> Optional argument BUFFER is the buffer to search for
> changes in."
>    (save-excursion
>      (if buffer (set-buffer buffer))
>      (let ((ol (semantic-overlays-in (max start
> (point-min))
> 				    (min end (point-max))))
> 	  (ret nil))
>        (while ol
> 	(when (semantic-overlay-get (car ol)
'semantic-change)
> 	  (setq ret (cons (car ol) ret)))
> 	(setq ol (cdr ol)))
>        (sort ret #'(lambda (a b) (<
> (semantic-overlay-start a)
> 				   (semantic-overlay-start b)))))))
> 
> I used `min' and `max', replaced 0 by `point-min', and
> removed the unnecessary
> `let' inside the `while' loop ;-)
> 
> If no objection, I will check the fix.
> 
> David
> 
> 
> 
> 
>
-------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Cedet-devel mailing list
> Cedet-devel@...
>
https://lists.sourceforge.net/lists/listinfo/cedet-devel

My computer is dead.  Would someone mind applying this
patch for me?  (Unless, of course, you can think of a
better way.) Thanks.

Eric

On 16 Sep 2003 12:34:14 -0400, Stuart Popejoy wrote:

> 
> [ ... ]
> 
> > Hmmm.  In our code, it says:
> > 
> >   (let ((changes-in-change
> (semantic-changes-in-region (1- start) (1+ end)))
> > 	)
> > 
> > without much testing to see if END was equal to
EOB. 
> Apparently this
> > is ok in GNU Emacs.  Did this happen editing the end
> of a file?
> > 
> > If so, does a patch like this help?
> > 
> > *** semantic-edit.el.~1.21.~	Tue Aug 26 10:25:43
2003
> > --- semantic-edit.el	Mon Sep 15 22:36:48 2003
> > ***************
> > *** 128,133 ****
> > --- 128,135 ----
> >   Optional argument BUFFER is the buffer to search
> for changes in."
> >     (save-excursion
> >       (if buffer (set-buffer buffer))
> > +     (if (< start 0) (setq start 0))
> > +     (if (> end (point-max)) (setq end
(point-max)))
> >       (let ((ol (semantic-overlays-in start end))
> >   	  (ret nil))
> >         (while ol
> > 
> > 
> 
> The patch works great. Thanks!
> 
> Stuart
> 
> 
> 
> 
>
-------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Cedet-devel mailing list
> Cedet-devel@...
>
https://lists.sourceforge.net/lists/listinfo/cedet-devel

[ ... ]

> Hmmm.  In our code, it says:
> 
>   (let ((changes-in-change (semantic-changes-in-region (1- start) (1+ end)))
> 	)
> 
> without much testing to see if END was equal to EOB.  Apparently this
> is ok in GNU Emacs.  Did this happen editing the end of a file?
> 
> If so, does a patch like this help?
> 
> *** semantic-edit.el.~1.21.~	Tue Aug 26 10:25:43 2003
> --- semantic-edit.el	Mon Sep 15 22:36:48 2003
> ***************
> *** 128,133 ****
> --- 128,135 ----
>   Optional argument BUFFER is the buffer to search for changes in."
>     (save-excursion
>       (if buffer (set-buffer buffer))
> +     (if (< start 0) (setq start 0))
> +     (if (> end (point-max)) (setq end (point-max)))
>       (let ((ol (semantic-overlays-in start end))
>   	  (ret nil))
>         (while ol
> 
> 

The patch works great. Thanks!

Stuart

On Tue, 16 Sep 2003 17:44:47 +0200 (CEST), David PONCE
wrote:

> 
> Eric,
> 
> Currently CEDET's package autoloads files don't have a
> consistent
> naming convention:
> 
> EDE      -> ede-loaddefs.el
> speedbar -> speedbar-defs.el
> EIEIO    -> eieio-defs.el
> semantic -> semantic-al.el
> cogre    -> cogre-defs.el
> 
> What do you think of adopting a common rule for naming
> autoloads
> files?
> 
> I propose to use names of the form
> <package-name>-loaddefs.el which
> seems consistent with what Emacs uses.  EDE already
> obeys to that
> convention, and if you agree with it I can try to do
> the necessary
> changes.  That shouldn't be too hard ;-)

Sounds good to me.

When I am able to compute again, I can make the
constructor for new autoload targets take the project
name and add loaddefs to the end.  The lack of this
feature added the problem.

> Another idea: maybe it could be worth generalizing the
> way semantic is
> loaded, and provide a <package>-load.el file for each
> CEDET package.
> Initially it would just load the package autoloads. 
In
> the future,
> it will simplify things, if we need to add some new
> options to a
> package.

I like that idea too.

> Finally, what do you think of removing the following
> files _that seems
> no more used_ from the repository:
> 
> semantic/semantic-autogen.el
> semantic/bovine/erlang.bnf

Good Idea

> semantic/wisent/Wisent.mk  (need update of
Project.ede)

Hmmm, I thought I had done that already, not that
I can actually go check. ;(

Eric

Eric,

Currently CEDET's package autoloads files don't have a consistent
naming convention:

EDE      -> ede-loaddefs.el
speedbar -> speedbar-defs.el
EIEIO    -> eieio-defs.el
semantic -> semantic-al.el
cogre    -> cogre-defs.el

What do you think of adopting a common rule for naming autoloads
files?

I propose to use names of the form <package-name>-loaddefs.el which
seems consistent with what Emacs uses.  EDE already obeys to that
convention, and if you agree with it I can try to do the necessary
changes.  That shouldn't be too hard ;-)

Using consistent names would allow to easily guess what autoloads
file is used for a package, and could simplify things in Makefiles
(see my comment in the main cedet/Makefile), and to setup CEDET.

Another idea: maybe it could be worth generalizing the way semantic is
loaded, and provide a <package>-load.el file for each CEDET package.
Initially it would just load the package autoloads.  In the future,
it will simplify things, if we need to add some new options to a
package.

Finally, what do you think of removing the following files _that seems
no more used_ from the repository:

semantic/semantic-autogen.el
semantic/bovine/erlang.bnf
semantic/wisent/Wisent.mk  (need update of Project.ede)

David

Hi Eric,

>   That's great.  A good way to avoid a `doit' type rule is to declare
>   a given rule as phony, like this:
[...]

I just used "doit" because the Make manual says:

-------------------
     clean: FORCE
             rm $(objects)
     FORCE:

...

   As you can see, using `FORCE' this way has the same results as using
`.PHONY: clean'.

   Using `.PHONY' is more explicit and more efficient.  However, other
versions of `make' do not support `.PHONY'; thus `FORCE' appears in
many makefiles.
-------------------

Probably, it isn't worth worrying about such compatibility issues
today ;-)

I will change to use .PHONY targets.

> Perhaps it would be better to have EDE create clean-grammars,
> clean-info, etc, as needed in each project, then this Makefile could
> have just:
[...]

That would be better.  More stuff to add to our TODO list ;-)

> In the meantime, I think this is a great place to start, and would be
> good for beta1.

Thanks!  Can I add it to the common package?  I will have to update
the Makefile to work from that directory too.

David

P.S.: Did you see my patch proposal to EDE:
<http://sf.net/mailarchive/forum.php?thread_id=3122800&forum_id=4869>

What do you think?