CEDET

Hi Martin,

> I'm using semantic 2.0 together with JDEE.
> It seems that semantic doesn't generate tags for anonymous java classes.
> Is there a way to turn this on?
> If not, I propose to consider this feature for future releases.
> An anonymous class is an inner class.

It is in my TODO list for the next release ;-)

David

You would probably have better luck using the latest cedet beta.  In
that version, the original .by file (replacement of the .bnf) is
stored so that the debugger can find it without having to do the more
complex setup described in 1.4.

In 1.4, your process sounds correct.  The end of a debug session
concludes by showing the BOVINATE buffer.  If the parser thinks there
is nothing to parse, then it would indeed skip debugging, and jump
right to that buffer.

This could happen if there is no parse table, or if the buffer
doesn't lex properly for some reason.  You can do this:

M-: (semantic-lex (point-min) (point-max)) RET

to see if that is doing ok.  (ie, does it think there is one big
comment?)

C-u M-x bovinate RET

will also show if the basics of parsing is working.

If that is the case and the debug is still not working, I have no
idea what else it might be.

Eric

>>> WZocher@... (Wolfgang Zocher) seems to think that:
>*,
>
>when I follow the instructions for debugging (Chapter 8 Debugging) language
>files using BNF with "bovinate-debug" I mostly have no success.
>
>Here is the scenario: I have a simple c-file hello.c, the c.bnf
>grammar-description as source and the corresponing lisp-table in semantic-c.el
>all in my current working directory. semantic is installed properly (I hope)
>with 
>
>        (setq semantic-load-turn-everything-on t)
>        (require 'semantic-load)
> 
>and semantic-load leaving untouched, i.e. the distributed version.
>
>Now I open two files; one buffer contains hello.c, another buffer
>semantic-c.el, I point to the table-structure in semantic-c.el and execute
> semantic-bovinate-debug-set-table. Then I set the pointer to the beginning of
>hello.c and execute bovinate-debug: the buffer containing semantic-c.el
>disappears and is replaced by *BOVINATE* which only contains one line with
>"nil". But sometimes, after endless trials, I get the layout described in the
>manual and the point now sits on the rule in the table corresponding to the
>token found in the c-source file and I can get through the source hitting the
>space-key ...
>
>Whats wrong with my setup? 
>What can I do to debug against the c.bnf file?
>
>Thanks in advance
>Wolfgang
>

-- 
          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

>>> Marcus Harnisch <marcus.harnisch@...> seems to think that:
>Hi,
>
>Is there any way for a single project to have multiple root
>directories? In a project I am working on, files that I change are in
>a separate tree. The union of both trees forms the actual project.
  [ ... ]

The answer is "yes" and "no".

The "no" just means there is no simple configuration for it.

For "yes", here is what you could do:

1) You could specify one project root, and add the second root to 
   include path for your language of choice.  Those files will then
   be pulled in.

2) Add a symbolic link to make it be in one project

3) Add one set of databases in one directory to
  `semanticdb-project-system-databases'.


I'll add a note for myself to make it a little easier than that.

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

>>> Igor Lautar <igor.lautar@...> seems to think that:
>Eric M. Ludlam <mailto:eric@...> wrote:
>
>> I've fixed semanticdb-mk.el, and attack a patch (against my
>> previously fixed version) below.
>
>semantic-cache is now created in every subdir.

Yay!

>>> How can emacs (semantic) be forced to read all semantic.cache files
>>> of the project? Can I list all semantic.cache files in .emacs ?
>> 
>> I have verified that this "works for me" in my C project "gnutalk".
>> (Also on sourceforge.)  The default value part had an issue fixed by
>> the second patch below.  My steps are below:
>> 
>> 1) In gnutalk/src/ run fixed semanticdb.sh on */*.c
>> 2) Start fresh emacs, load gtalkd/gtd_ann.c
>> 3) go to a call to HOST_somethingidontremember
>> 4) C-c , J
>> 
>> At this point, the default is where I go and fixed semantic-complete
>> finds it in the correct buffer.  If I type it, the conveniences there
>> also work on that symbol found in a different directory.
>
>this still does not work as I expect, only tags from semantic.cache which
>contains opened file are displayed (and searched).
>
>My test:
>- root_dir
>  .
>  .
>  .
>  - dir
>    .
>    .
>    .
>    - dir1
>     .
>     .
>     .
>     -file1
>     -file2
>     -semantic.cache
>   - dir2
>     .
>     .
>     .
>     -file3 (contains tag scsi_write)
>     -file4
>     -semantic.cache
>
>semanticdb-project-roots: root_dir
>- open file1 in emacs, C-c , J and it doesn't display scsi_write tag
>
>can cache files be preloaded at emacs startup?

Yes.  You can add something like:

(semanticdb-load-database "/blah/blah/blah/semantic.cache")

>this is quite a complex project, the root_dir has around 2 MLOC.
>can this be a problem?
  [ ... ]

I've added this to semanticdb-find to help debug things.  Perhaps
this will also help you analyze closer if it is a path issue, or
something else.


(defun semanticdb-find-test-translate-path (&optional arg)
  "Call and output results of `semanticdb-find-translate-path'"
  (interactive "P")
  (let ((p (semanticdb-find-translate-path nil arg)))
    ;; Output the result
    (with-output-to-temp-buffer "*Translated Path*"
      (while p
	(princ (semanticdb-printable-name (car p)))
	(princ "\n")
	(setq p (cdr p)))
      )
    (message "%d paths found." (length p))
    ))

M-x semanticdb-find-test-translate-path RET

will be what tooltips and such search.

C-u M-x semanticdb-find-test-translate-path RET

will be what the jump function you are using would specify.

You can compare the output to what you think should be in the list.

Eric

Markus Gritsch wrote:
> 
> The code in linemark.el prevents from setting bookmarks in buffers which 
> do not contain a file.  I find the bookmark feature very useful, also in 
> *compilation* and other non-file buffers, therefore I hacked a bit in 
> linemark.el to allow this.  I don't know if there are cases where the 
> modifications don't work, but up to know I didn't encounter any problem 
> with it.

There was a problem when calling (lmcompile-do-highlight) which is fixed 
now.

Please find the new diff attached.

Have fun,
Markus

Thanks,  thats a good idea.  I'll merge your change into soon.

Eric

>>> Markus Gritsch <gritsch@...> seems to think that:
>Markus Gritsch wrote:
>> 
>> The code in linemark.el prevents from setting bookmarks in buffers which 
>> do not contain a file.  I find the bookmark feature very useful, also in 
>> *compilation* and other non-file buffers, therefore I hacked a bit in 
>> linemark.el to allow this.  I don't know if there are cases where the 
>> modifications don't work, but up to know I didn't encounter any problem 
>> with it.
>
>There was a problem when calling (lmcompile-do-highlight) which is fixed 
>now.
>
>Please find the new diff attached.
  [ ... ]

-- 
          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

>>> Igor Lautar <igor.lautar@...> seems to think that:
>Eric M. Ludlam <mailto:eric@...> wrote:
><cut>
>
>> Hi,
>> 
>>   Here is a patch to sematnicdb-mk.el that should get you going.
>> There seemed to be something strange about how the load-file-path was
>> calculated on my version of Emacs.  Hopefully it is safer now to.
>> 
>>   If needed, you can download the file from CVS directly instead.
>
>I've applied the patch and it works, partially. Example:
>I have an dir with some dirs that contain source files.
>--dir
>   |-subdir1
>   |   |-files1
>   |-subdir2
>   |   |-files2
>and so on...
>
>I do the following in dir:
>find . -name '*.[c|h]' | xargs > file_list
>
>then emacs <parms to launch semanticdb-mk.el> <contents of file_list>
>
>it works ok, loads all files listed but semantic.cache is not created for
>every subdir.

I've managed to verify this behavior.  It seems only the very first
file in the list gets the correct answer when subdirectories are
involved.  If I CD into each directory and re-run, it seems to get
the right answer.

I've fixed semanticdb-mk.el, and attack a patch (against my
previously fixed version) below.

>also, if I create semantic.cache for subdir1 and subdir2,
>and semanticdb-project-roots is dir, then open file in subdir2,
>I cannot semantic-complete-jump to tag in subdir1.
>
>How can emacs (semantic) be forced to read all semantic.cache files
>of the project? Can I list all semantic.cache files in .emacs ?

I have verified that this "works for me" in my C project "gnutalk".
(Also on sourceforge.)  The default value part had an issue fixed by
the second patch below.  My steps are below:

1) In gnutalk/src/ run fixed semanticdb.sh on */*.c
2) Start fresh emacs, load gtalkd/gtd_ann.c
3) go to a call to HOST_somethingidontremember
4) C-c , J

At this point, the default is where I go and fixed semantic-complete
finds it in the correct buffer.  If I type it, the conveniences there
also work on that symbol found in a different directory.

My sample (now that I have one ready to go to test against) seems to
have issues with the summary/autocompletion mode that I still need to
look into.  Strangely, this feature works on what I do at work which
is much more complex.  I still need to look into that.

>I'm aware semantic searches one level into include files, can
>this be increased?

Yes.  See semanticdb-find.el, around line 237.  If anyone wants to
figure out a good way to fix it, that would be grand.

Eric

-----------
*** semanticdb-mk.el.~1.2.~	2004-08-02 21:38:18.000000000 -0400
--- semanticdb-mk.el	2004-08-08 22:21:35.000000000 -0400
***************
*** 52,58 ****
    (while (and args
  	      (not
  	       (progn
! 		 (message "Compare %s to %s" (car args) "-l")
  		 (string= (car args) "-l"))))
      (setq args (cdr args)))
    (when args
--- 52,58 ----
    (while (and args
  	      (not
  	       (progn
! 		 ;(message "Compare %s to %s" (car args) "-l")
  		 (string= (car args) "-l"))))
      (setq args (cdr args)))
    (when args
***************
*** 60,69 ****
      ;; Grab the rest of the file names.
      ;; For each file, load it in, let semantic evaluate it for tags.
      (while args
!       (message "Loading %s" (car args))
!       (find-file (car args))
!       (semantic-fetch-tags)
!       (setq args (cdr args)))
      ))
  
  ;; Save the databases.
--- 60,74 ----
      ;; Grab the rest of the file names.
      ;; For each file, load it in, let semantic evaluate it for tags.
      (while args
!       (princ (concat "Loading " (car args) "... "))
!       (save-window-excursion
! 	(let* ((buffer (find-file-noselect (car args)))
! 	       (tags nil))
! 	  (set-buffer buffer)
! 	  (setq tags (semantic-fetch-tags))
! 	  (princ (length tags))
! 	  (princ " tags found .\n"))
! 	(setq args (cdr args))))
      ))
  
  ;; Save the databases.


-----------
*** semantic-complete.el.~1.35.~	2004-08-02 21:46:04.000000000 -0400
--- semantic-complete.el	2004-08-08 22:33:49.000000000 -0400
***************
*** 302,308 ****
  		   semantic-completion-collector-engine)))
  	  (when ml
  	    ;; We don't care about uniqueness.  Just guess for convenience
! 	    (setq tag (car (semanticdb-find-result-nth ml 0)))))
  	;; save it
  	(setq semantic-complete-active-default tag)
  	;; Return it.. .whatever it may be
--- 302,308 ----
  		   semantic-completion-collector-engine)))
  	  (when ml
  	    ;; We don't care about uniqueness.  Just guess for convenience
! 	    (setq tag (semanticdb-find-result-nth-in-buffer ml 0))))
  	;; save it
  	(setq semantic-complete-active-default tag)
  	;; Return it.. .whatever it may be

-- 
          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,

  speedbar.el needs to be loaded after the cedet load file.  This
allows the Speedbar in CEDET to be loaded in front of the one that is
a part of Emacs.

  You can also just delete the speedbar from your Emacs distribution.

Eric

>>> "luc strijbosch" <strijbozz@...> seems to think that:
>Hi,
>
>I compiled cedet1.0beta3b succesfully but when I load emacs the following 
>error shows up :
>
>error: "Package speedbar does not define speedbar-version"
>
>I believe it has something to do with the inversion module looking at the 
>backtrace shown below.
>
>Does anyone have a clue how to solve the problem ? For example 
>complete-ia-symbol doesn't work and reports me the error show above, so the 
>problem has to be solved for the package to work properly.
>
>Thanx in advance,
>Luc
>
>
>---- Back trace ------
>
>Debugger entered--Lisp error: (error "Package speedbar does not define 
>speedbar-version")
>  signal(error ("Package speedbar does not define speedbar-version"))
>  error("Package %S does not define %S-version" speedbar speedbar)
>  inversion-package-version(speedbar)
>  inversion-test(speedbar "0.15beta1")
>  inversion-require(speedbar "0.15beta1")
>  
>byte-code("&#1102;&#1072;!&#9516;&#1102;&#1073;!&#9516;&#1102;&#1094;!&#9516;&#1076;&#1077;&#1092;\"&#9508;" 
>[require semantic semantic-util inversion inversion-require speedbar 
>"0.15beta1"] 3)
>  require(semantic-sb)
>  
>byte-code("&#1102;&#1072;!&#9516;&#1102;&#1073;!&#9516;&#1094;&#1076;&#1077;&#9616;&#9516;&#1092;&#1075;!&#9488;&#1102;&#1093;!&#9516;&#9484;&#1080;&#1081;&#1082;\"&#9516;&#1083;&#1084;&#1094;&#1085;&#1086;&#1072;&#1086;&#1087;&&#9508;" 
>[require semantic semantic-format nil (byte-code 
>"&#1102;&#1072;!&#9516;&#1073;&#1094;!&#9488;&#1076;&#1077;&#1092;&#1075;&#1093;$&#9516;&#1080;&#1077;&#1093;\"&#9516;&#1093;&#9508;" 
>[require imenu featurep xemacs ad-add-advice imenu--create-menu-1 ... around 
>nil ad-activate] 5) ((error)) featurep speedbar semantic-sb add-hook 
>speedbar-load-hook #[nil "&#1102;&#1072;!&#9508;" [require semantic-sb] 2] 
>custom-declare-group semantic-imenu "Parser Generator Imenu interface." 
>:group imenu] 8)
>  require(semantic-imenu)
>  
>byte-code("&#1102;&#1072;!&#9516;&#1102;&#1073;!&#9516;&#1102;&#1094;!&#9516;&#1076;&#1077;&#1092;&#1075;&#1093;&#1072;%&#9508;" 
>[require semantic semantic-ctxt semantic-imenu custom-declare-group senator 
>nil "SEmantic NAvigaTOR." :group] 6)
>  global-senator-minor-mode(1)
>  semantic-load-enable-code-helpers()
>  (if (string-match "XEmacs\\|Lucid" emacs-version) (progn (if ... ...) 
>(setq options-save-faces t) (if ... ...)) (if (file-readable-p 
>"~/.gnu-emacs") (load "~/.gnu-emacs" nil t) (if ... ...)) (tool-bar-mode -1) 
>(scroll-bar-mode -1) (global-font-lock-mode t) (show-paren-mode t) 
>(add-to-list (quote load-path) "~/elisp") (require (quote color-theme)) 
>(color-theme-jb-simple) (load-library "php-mode-102") (add-hook (quote 
>php-mode-user-hook) (quote turn-on-font-lock)) (load-library "setnu") 
>(load-library "setnu+") (setq-default truncate-lines t) (defun dos2unix nil 
>"Convert a buffer from MS-DOS ^M end of lines to UNIX end of lines." 
>(interactive) (goto-char ...) (while ... ...)) (defun unix2dos nil "Convert 
>a buffer from UNIX end of lines to MS-DOS ^M end of lines." (interactive) 
>(goto-char ...) (while ... ...)) (autoload (quote expand-member-functions) 
>"member-functions" "Expand C++ member function declarations" t) (add-hook 
>(quote c++-mode-hook) (lambda nil ...)) (global-set-key "" (quote 
>goto-line)) (setq visible-bell t) (fset (quote yes-or-no-p) (quote 
>y-or-n-p)) (require (quote bs)) (global-set-key "" (quote bs-show)) (setq 
>scroll-step 1) (defun scroll-one-line-up (&optional arg) "Scroll the 
>selected window up (forward in the text) one line (or N lines)." 
>(interactive "p") (scroll-up ...)) (defun scroll-one-line-down (&optional 
>arg) "Scroll the selected window down (backward in the text) one line (or 
>N)." (interactive "p") (scroll-down ...)) (global-set-key [M-down] (quote 
>scroll-one-line-up)) (global-set-key [M-up] (quote scroll-one-line-down)) 
>(mwheel-install) (global-set-key [C-next] (quote scroll-other-window)) 
>(global-set-key [C-prior] (quote scroll-other-window-down)) (autoload (quote 
>iswitchb) "iswitchb" "Run iswitchb" t) (setq read-buffer-function (quote 
>iswitchb-read-buffer)) (setq iswitchb-default-method (quote samewindow)) 
>(iswitchb-default-keybindings) (setq next-line-add-newlines nil) (defun 
>ascii-table nil "Print the ascii table. Based on a defun by Alex Schroeder 
><asc@...>" (interactive) (switch-to-buffer "*ASCII*") (erase-buffer) 
>(insert ...) (let ... ...) (beginning-of-buffer)) (transient-mark-mode t) 
>(require (quote toggle-buffer)) (require (quote ibs)) (global-set-key [f1] 
>(quote find-file)) (global-set-key [f2] (quote kill-buffer)) (global-set-key 
>[f3] (quote electric-buffer-list)) (global-set-key [f4] (quote shell)) 
>(global-set-key [f5] (quote compile)) (global-set-key [f7] (quote gdb)) 
>(global-set-key [f9] (quote delete-other-windows)) (global-set-key [f10] 
>(quote joc-toggle-buffer)) ...)
>  eval-buffer(#<buffer  *load*> nil "~/.emacs" nil t)
>  load-with-code-conversion("/home/luc/.emacs" "~/.emacs" t t)
>  load("~/.emacs" t t)
>  #[nil 
>"&#9496;&#8776;	&#1092;=&#9488;&#1075;&#1093;Q&#9484;5	&#1080;=&#9488;'&#1081;&#1075;&#1082;&#1083;#&#9488;#&#1084;&#9484;5&#1085;&#9484;5	&#1086;=&#9488;1&#1087;&#9484;5&#1075;&#1103;Q&#1088;&#1089;\n&#1088;&#9524;#&#9516;&#1088;=&#9488;F&#1082;&#9488;&#9516;&#1090;!&#1091; 
>&#9488;&#9516;&#1078;!&#9524;&#1074;P&#1100;
>!&#9488;d
>&#9484;o&#1100;\f!&#9488;n\f&#9484;o&#1082;&#9524;&#9488;&#9508;&#1099;\f\"&#9488;&#9496;&#1079;&#1096;\f#&#9516;&#1101;&#1097;!&#9516;\f*&#9500;&#8730;&#1082; 
>&#1089;&#1095;&#1088;&#9524;#))&#9508;" [init-file-user system-type 
>user-init-file-1 user-init-file source alt ms-dos "~" "/_emacs" windows-nt 
>directory-files nil "^\\.emacs\\(\\.elc?\\)?$" "~/.emacs" "~/_emacs" vax-vms 
>"sys$login:.emacs" "/.emacs" t load file-name-extension "elc" 
>file-name-sans-extension ".el" file-exists-p file-newer-than-file-p message 
>"Warning: %s is newer than %s" sit-for 1 "default" inhibit-default-init 
>inhibit-startup-message] 5]()
>  command-line()
>  normal-top-level()
>
>----- End of trace ---
  [ ... ]

Eric M. Ludlam wrote:

 > >* Unfortunately I cannot use the new CVS Emacs from
 > >http://www.crasseux.com/emacs/ because (lmcompile-do-highlight) needs a
 > >replacement for `compile-reinitialize-errors.  At leased it is a known
 > >issue due to the comment in lmcompile.el.
 > >
 > >* As suggested in a previously posted patch lmcompile.el line 164 
should
 > >use 'linemark-go-face instead of 'linemark-funny-face to highlight the
 > >matches, because the funny face is already used by bookmarks.
 >
 > Thanks for the reminder.  I put it in this time.  I also found why
 > lmcompile doesn't work with my CVS Emacs, which I think is too old for
 > your previously mentioned problem.

So do you have a version of lmcompile.el which works with the current 
CVS Emacs?

 > The semantic idle handling code tries to be smart about when
 > debug-on-error is on.  I've just added this patch to my local copy
 > which should then also guarantee that C-g is enabled.  Do you want to
 > try it?

No thanks, I think my 'hanging'-problem is related to something else 
(see below), because 1) I have set debug-on-error to nil and 2) I can 
successfully interrupt the hanging by your suggested C-g.

 > I haven't had the semantic parser lock me out in a very long time,
 > though it could be the language you are using.  Is it related to a
 > particular file?

Maybe it's indeed related to a particular language.  I had the hanging 
when trying out if the reversed methods order patch in Python classes 
was applied in the beta3, and now I can reliably reproduce the hanging:

I tested it using GNU Emacs from http://www.crasseux.com/emacs/ dating 
from 2003-12-28 on Windows XP.

*) start Emacs
*) C-x C-f test.py <RET>
*) type in the following code:

class C:
     def

and wait some seconds -> 100% CPU probably due to the incomplete Python 
code.


Kind regards,
Markus

>>> Markus Gritsch <gritsch@...> seems to think that:
>Eric M. Ludlam wrote:
>
> > >* Unfortunately I cannot use the new CVS Emacs from
> > >http://www.crasseux.com/emacs/ because (lmcompile-do-highlight) needs a
> > >replacement for `compile-reinitialize-errors.  At leased it is a known
> > >issue due to the comment in lmcompile.el.
> > >
> > >* As suggested in a previously posted patch lmcompile.el line 164 
>should
> > >use 'linemark-go-face instead of 'linemark-funny-face to highlight the
> > >matches, because the funny face is already used by bookmarks.
> >
> > Thanks for the reminder.  I put it in this time.  I also found why
> > lmcompile doesn't work with my CVS Emacs, which I think is too old for
> > your previously mentioned problem.
>
>So do you have a version of lmcompile.el which works with the current 
>CVS Emacs?

My emacs is from Jan 22, 2004, so it is not the latest and greatest,
and as such, I just don't know.

> > The semantic idle handling code tries to be smart about when
> > debug-on-error is on.  I've just added this patch to my local copy
> > which should then also guarantee that C-g is enabled.  Do you want to
> > try it?
>
>No thanks, I think my 'hanging'-problem is related to something else 
>(see below), because 1) I have set debug-on-error to nil and 2) I can 
>successfully interrupt the hanging by your suggested C-g.
>
> > I haven't had the semantic parser lock me out in a very long time,
> > though it could be the language you are using.  Is it related to a
> > particular file?
>
>Maybe it's indeed related to a particular language.  I had the hanging 
>when trying out if the reversed methods order patch in Python classes 
>was applied in the beta3, and now I can reliably reproduce the hanging:
>
>I tested it using GNU Emacs from http://www.crasseux.com/emacs/ dating 
>from 2003-12-28 on Windows XP.
>
>*) start Emacs
>*) C-x C-f test.py <RET>
>*) type in the following code:
>
>class C:
>     def
  [ ... ]

The first problem is in the lexical analyzer.  It gets confused here,
and I have the following patch.

Getting past the lexer leads to a hang in wisent.  Perhaps David has
a notion of how to handle that part.

Eric

--------------
*** wisent-python.el.~1.44.~	2004-06-29 22:57:01.000000000 -0400
--- wisent-python.el	2004-08-03 22:32:00.000000000 -0400
***************
*** 1,6 ****
  ;;; wisent-python.el --- LALR grammar for Python
  ;;
! ;; Copyright (C) 2002 Richard Kim
  ;;
  ;; Author: Richard Kim <ryk@...>
  ;; Maintainer: Richard Kim <ryk@...>
--- 1,6 ----
  ;;; wisent-python.el --- LALR grammar for Python
  ;;
! ;; Copyright (C) 2002, 2004 Richard Kim
  ;;
  ;; Author: Richard Kim <ryk@...>
  ;; Maintainer: Richard Kim <ryk@...>
***************
*** 263,271 ****
            (error "You have found a bug in python-next-line")))
      ;; the point now should be at the end of a line
      (forward-line 1)
!     (while (and (looking-at "\\s-*\\(\\s<\\|$\\)")
!                 (not (eobp))) ;; skip blank and comment lines
!       (forward-line 1))))
  
  (defun python-scan-lists ( &optional target-column )
    "Without actually changing the position, return the buffer position of
--- 263,272 ----
            (error "You have found a bug in python-next-line")))
      ;; the point now should be at the end of a line
      (forward-line 1)
!     (while (looking-at "\\s-*\\(\\s<\\|$\\)")  ;; skip blank and comment lines
!       (forward-line 1)
!       (if (eobp) (error "Unterminated List."))
!       )))
  
  (defun python-scan-lists ( &optional target-column )
    "Without actually changing the position, return the buffer position of

-- 
          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

>>> langrind@... seems to think that:
>Hi Y'all,
>I have a class defined in a header file, and the name of the header is the same as the name of the class - e.g. class Foo is defined in Foo.h.
>If I am in some other file, and I have a declaration:
>  Foo foo;
>And I place the mark at Foo, the class is shown below the mode line, so semantic knows the tag.
>I invoke semantic-complete-jump, and type Foo<return> and the minibuffer says "[Not Unique]"
>If I then type a space, the string in the minibuffer is partially completed, such that it says:
>   Foo.
>If I type another space, it further completes to
>   Foo.h
>Thus it is possible to visit the lines that say "#include <Foo.h>", which is a pretty neat feature, but it is not possible to visit the class definition of Foo.
>Is there a function that behaves just like semantic-complete-jump, but without the completion feature? Or is there some other way to deal with this problem.
>Thanks, Nik

Hi,

  The jump routine you used (C-c , J) will jump to remote files as you
expect, but it includes matching header files and the like.  It only
knows about files you have visited, so if you haven't visited Foo.c,
it may not know about it.

  Also, SPC will complete over word boundaries ("." "_" etc) and since
Foo.h and Foo are similar, SPC will add the ., but TAB will show
completions.

  The purpose of the completion is to differentiate between multiple
symbols with the same name, such as a prototype, and the actual
definition.  Pressing TAB will first show you the list of all
possible symbols (use TAB instead of SPACE for some operations.)
Hitting tab multiple times will highlight the definition you could
jump to, thus it could be that:

C-c , J FOO TAB
% shows list
TAB
% highlight foo.h
TAB
% hilight Foo in header file
TAB
% hilight Foo in C file
RET
% go there

Placing the cursor on the symbol "Foo" in your buffer should also
offer up a default, but that has proven unreliable in the past.

Lastly, if your files are in different directories, you neet to set
up the variable `semanticdb-project-roots'.

Good Luck
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

>>> Igor Lautar <igor.lautar@...> seems to think that:
>hi,
>
>I'm using semantic from cedet-1.0beta3a with GNU emacs 21.3.1 on windows 2k
>SP4.
>(all patches installed)
>
>cut from semanticdb.sh:
>
>exec $emacs -batch -l "${loadpath}/semanticdb-mk.el" $files 
>
>I've tried the following:
>
>bash-2.05b$ find . -name '*.[c|h]' | /cygdrive/c/Program\
>Files/emacs-21.3/bin/emacs.exe -L ~/.emacs.d/cedet-1.0beta3a/common/" -batch
>-l "~/.emacs.d/cedet-1.0beta3a/semantic/semanticdb-mk.el" -
>
>But emacs eats 100% CPU and does nothing. Where could be the problem ?
>
>The same thing is if I just give one file to parse (i.e. not with find).
  [ ... ]

Hi,

  Here is a patch to sematnicdb-mk.el that should get you going.
There seemed to be something strange about how the load-file-path was
calculated on my version of Emacs.  Hopefully it is safer now to.

  If needed, you can download the file from CVS directly instead.

Eric

---------
*** semanticdb-mk.el.~1.1.~	2002-08-10 21:47:36.000000000 -0400
--- semanticdb-mk.el	2004-08-02 21:35:28.000000000 -0400
***************
*** 1,6 ****
  ;;; semanticdb-mk.el --- Command line database builder
  
! ;;; Copyright (C) 2002 Eric M. Ludlam
  
  ;; Author: Eric M. Ludlam <zappo@...>
  ;; Keywords: tags
--- 1,6 ----
  ;;; semanticdb-mk.el --- Command line database builder
  
! ;;; Copyright (C) 2002, 2004 Eric M. Ludlam
  
  ;; Author: Eric M. Ludlam <zappo@...>
  ;; Keywords: tags
***************
*** 30,88 ****
  
  ;;; Code
  ;;
- (defun semanticdb-mk-find-eieio ()
-   "Find EIEIO in the usual places."
-   (cond 
-    ((locate-library "eieio")
-     nil)
-    ((getenv "EIEIO_LOADPATH")
-     (add-to-list 'load-path (getenv "EIEIO_LOADPATH")))
-    (t;; For now, lets hope it was installed in their .emacs file.
-     (let ((de (find-file-noselect "~/.emacs"))
- 	  (path nil))
-       (save-excursion
- 	(set-buffer de)
- 	(goto-char (point-min))
- 	(while (not path)
- 	  (re-search-forward "eieio")
- 	  (if (re-search-backward "\"" nil t)
- 	      (progn
- 		(mark-sexp 1)
- 		(setq path
- 		      (buffer-substring-no-properties (point) (mark))))))
- 	(add-to-list 'load-path (read path)))))))
- 
  (if (not noninteractive)
      (error "You should not load semanticdb-mk interactivly."))
  
! ;; Force semantic onto the load path
! (add-to-list 'load-path (file-name-directory load-file-name))
  
! ;; Add eieio to the path.
! (semanticdb-mk-find-eieio)
  
! ;; Initialize semantic
! (require 'semantic-load)
  
  ;; Turn on semanticdb
  (global-semanticdb-minor-mode 1)
  
  ;; Process loaded buffers from the command line.
  (let ((args command-line-args))
    ;; Move past this load file being loaded.
!   (while (not
! 	  (progn
! 	    ;;(message "Compare %s to %s" (car args) load-file-name)
! 	    (string= (car args) load-file-name)))
!     (setq args (cdr args)))
!   (setq args (cdr args))
!   ;; Grab the rest of the file names.
!   ;; For each file, load it in, let semantic evaluate it for tags.
!   (while args
!     (message "Loading %s" (car args))
!     (find-file (car args))
      (setq args (cdr args)))
!   )
  
  ;; Save the databases.
  (semanticdb-save-all-db)
--- 30,73 ----
  
  ;;; Code
  ;;
  (if (not noninteractive)
      (error "You should not load semanticdb-mk interactivly."))
  
! ;; Find our source directory
! (let (fname semanticdir cedetdir loadfile)
  
!   (setq fname load-file-name)
!   (setq semanticdir (file-name-directory fname))
!   (setq cedetdir (expand-file-name (concat semanticdir "../common/")))
!   (setq loadfile (concat cedetdir "cedet.el"))
  
!   (load-file loadfile))
  
  ;; Turn on semanticdb
+ (message "Enabling semanticdb...")
  (global-semanticdb-minor-mode 1)
+ (message "Enabling semanticdb...done")
  
  ;; Process loaded buffers from the command line.
  (let ((args command-line-args))
    ;; Move past this load file being loaded.
!   (while (and args
! 	      (not
! 	       (progn
! 		 (message "Compare %s to %s" (car args) "-l")
! 		 (string= (car args) "-l"))))
      (setq args (cdr args)))
!   (when args
!     (setq args (cdr (cdr args)))
!     (message "Args left: %S" args)
!     ;; Grab the rest of the file names.
!     ;; For each file, load it in, let semantic evaluate it for tags.
!     (while args
!       (message "Loading %s" (car args))
!       (find-file (car args))
!       (semantic-fetch-tags)
!       (setq args (cdr args)))
!     ))
  
  ;; Save the databases.
  (semanticdb-save-all-db)