CEDET

>>> kin@... seems to think that:
>Hi,
>
>I'm a new user of semantic.  Using semantic for parsing,
>I've been rolling my own intellisense functionality and am
>happy with the results so far.
>
>However, a number of our headers contain structs which
>contain function pointer decl that semantic seems to have
>problem parsing.  For example, this struct:
>
>struct foo {
>	int (*bar) (void *);
>};
>
>Gives this parse tree:
>
>(("Types"
>  ("foo" type "struct" nil nil nil nil nil #<overlay in no buffer>)))
>
>If I change 
>	int (*bar) (void *);
>to
>	foobar (*bar) (void *);
>
>I got this parse tree instead:
>
>(("Types"
>  ("foo" type "struct"
>   (("foobar" function "int"
>     (("" variable
>       ("bar" type "class")
>       nil nil nil nil #<overlay in no buffer>))
>     nil nil
>     ((reparse-symbol . classsubparts))
>     #<overlay in no buffer>))
>   nil nil nil nil #<overlay in no buffer>)))
>
>Which is still a problem, since "bar" should be a "variable"
>in "foo".
  [ ... ]

Hi,

  You are correct.  Supporting function pointers has been on the TODO
  list for the C parser for quite some time.

  I've checked into CVS a c.by and semantic-c.el with rudimentary
  function pointer support in the parser, and in the output format
  functions.

  If anyone is interested in porting that back to the 1.4
  branch please do.  I looked into it briefly but found what looked
  like some bad merge problems.

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

On Mon, 26 May 2003 14:48:55 -0600, Jan Sacharuk <jan@...> wrote:

Slight correction. I checked, and according to 'customize', they ARE off, 
despite the fact that both of them are currently actively showing that 
they're running. By toggling the state on then off and setting it and 
saving it, things properly turn off.

JS

> I can't get them to stay turned off. I explicitly set
>
> (global-semantic-summary-mode nil)
> (global-semantic-stickyfunc-mode nil)
>
> near the bottom of my _emacs file, after everything has been activated, 
> and still no luck. As well, I've SEVERAL times set them both to be off in 
> the customizations, and made sure that the settings were saved for future 
> sessions, but no matter what, they keep coming back, and I have to turn 
> them off manually.
>
> Any help?
>
> JS
>



-- 
Jan Sacharuk
Programmer
BioWare Corp.
jan@...

You may find it easier to not set the variable
"semantic-load-turn-everything-on' or
`semantic-load-turn-useful-things-on' to t, but instead activate each
individual tool you actually want to use instead.

I think David Ponce may know more about how the customize stuff works
with the minor modes.

Eric

>>> Jan Sacharuk <jan@...> seems to think that:
>On Mon, 26 May 2003 14:48:55 -0600, Jan Sacharuk <jan@...> wrote:
>
>Slight correction. I checked, and according to 'customize', they ARE off, 
>despite the fact that both of them are currently actively showing that 
>they're running. By toggling the state on then off and setting it and 
>saving it, things properly turn off.
>
>JS
>
>> I can't get them to stay turned off. I explicitly set
>>
>> (global-semantic-summary-mode nil)
>> (global-semantic-stickyfunc-mode nil)
>>
>> near the bottom of my _emacs file, after everything has been activated, 
>> and still no luck. As well, I've SEVERAL times set them both to be off in 
>> the customizations, and made sure that the settings were saved for future 
>> sessions, but no matter what, they keep coming back, and I have to turn 
>> them off manually.
>>
>> Any help?
>>
>> JS
>>
>
>
>

-- 
          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

semantic has adapted eldoc to also display C/C++ summaries.

Here is an example of what you want.

(defsubst senator-find-current-symbol-tag (sym)
  "Search for a semantic tag with name SYM.
Return the tag found or nil if not found."
  (car (if (and (featurep 'semanticdb) semanticdb-current-database)
           (cdar (semanticdb-deep-find-tags-by-name sym))
         (semantic-deep-find-tags-by-name sym (current-buffer)))))

(defun senator-eldoc-print-current-symbol-info-default ()
  "Return a string message describing the current context."
  (let (sym found)
    (and
     ;; 1- Look for a tag with current symbol name
     (setq sym (car (semantic-ctxt-current-symbol)))
     (not (setq found (senator-find-current-symbol-tag sym)))
     ;; 2- Look for a keyword with that name
     (semantic-lex-keyword-p sym)
     (not (setq found (semantic-lex-keyword-get sym 'summary)))
     ;; 3- Look for a tag with current function name
     (setq sym (car (semantic-ctxt-current-function)))
     (not (setq found (senator-find-current-symbol-tag sym)))
     ;; 4- Look for a keyword with that name
     (semantic-lex-keyword-p sym)
     (setq found (semantic-lex-keyword-get sym 'summary)))
    found))

(defun senator-eldoc-print-current-symbol-info ()
  "Print information using `eldoc-message' while in function `eldoc-mode'.
You can override the info collecting part with `eldoc-current-symbol-info'."
  (let* ((s (semantic-fetch-overload 'eldoc-current-symbol-info))
         found)
    
    (if s (setq found (funcall s))
      (setq found (senator-eldoc-print-current-symbol-info-default)))
    
    (eldoc-message
     (cond ((stringp found)
            found)
           ((semantic-tag-p found)
            (semantic-format-tag-summarize found nil senator-eldoc-use-color))
           (t nil)
           ))))

(defadvice eldoc-print-current-symbol-info (around senator activate)
  "Enable ELDOC in non Emacs Lisp, but semantic-enabled modes."
  (if (eq major-mode 'emacs-lisp-mode)
      ad-do-it
    (if (semantic-active-p)
        (if (eldoc-display-message-p)
            (senator-eldoc-print-current-symbol-info))
      (eldoc-mode -1))))

Eric

>>> henrik.jonsson@... seems to think that:
>I tried invoking global-semantic-summary-mode. This seems to invoke eldoc as
>well. I'm using C/C++, I think that eldoc only is for lisp. I didn't get much
>wiser reading the doc strings in semantic-utils-modes.el.
>
>Could someone give some small snippet including semantic-ctxt-current-symbol. I
>can't get it to work.
>
>/henrik
>
>
>
>
>
>"Eric M. Ludlam" <eric@...> on 2003-05-22 17:05:37
>
>
>To:   Henrik Jonsson/SE/Transport/Bombardier@...
>cc:   cedet-semantic@...
>
>Security Level:?         Internal
>
>Subject: Re[1]: [cedet-semantic] Question on the se-show example
>
>
>
>Does `global-semantic-summary-mode' do what you want?
>
>You may find a long list of useful minor modes in
>semantic-util-modes.el
>
>To answer your more specific question, you can use
>`semantic-ctxt-current-symbol' to get the text from the buffer which
>you can then use to lookup your symbol.
>
>Eric
>
>>>> henrik.jonsson@... seems to think that:
>>Hi!
>>
>>I'm a semantic newcomer. I would like to have a function that displays the
>>arguments to the function that I am about to call in the message area. I found
>>se-show in the semantic-example.el file and it almost do what I want. If I call
>>it it asks for a symbol name, however the default value is already inserted, so
>>I only need to press enter to see a beatifull function argument list in the
>>message area.
>>
>>How can I avoid the question of the symbol and only let it read my newly
>written
>>function name?
>>It sounds simple, and I have tried to do some changes without success.
>>
>>(defun se-show (name)
>>  "Show important information about some token with NAME.
>>Shows how to:
>> * Create a variable with `semantic-token->text-functions'.
>> * Create the colorized text."
>>  (interactive (list (semantic-read-symbol "Symbol: ")))
>>  (let ((tok (semantic-find-nonterminal-by-name name (current-buffer) t))
>>     parent
>>     msg)
>>    (if tok
>>     (save-excursion
>>       ;; Token text functions can take a PARENT argument.  To find this
>>       ;; really quickly, jump to TOK, then use the overlay mechanism to
>>       ;; Find it.
>>       (goto-char (semantic-token-start tok))
>>       (setq parent (semantic-current-nonterminal-parent))
>>       ;; Create a message
>>       (setq msg (funcall se-summary-function tok parent t))
>>       ;; Display it
>>       (message msg)))))
>>
>>
>>Regards
>>
>>Henrik
>>
>>
>>
>>
>>-------------------------------------------------------
>>This SF.net email is sponsored by: ObjectStore.
>>If flattening out C++ or Java code to make your application fit in a
>>relational database is painful, don't do it! Check out ObjectStore.
>>Now part of Progress Software. http://www.objectstore.net/sourceforge
>>_______________________________________________
>>cedet-semantic mailing list
>>cedet-semantic@...
>>https://lists.sourceforge.net/lists/listinfo/cedet-semantic
>>
>
>--
>          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
>
>
>-------------------------------------------------------
>This SF.net email is sponsored by: ObjectStore.
>If flattening out C++ or Java code to make your application fit in a
>relational database is painful, don't do it! Check out ObjectStore.
>Now part of Progress Software. http://www.objectstore.net/sourceforge
>_______________________________________________
>cedet-semantic mailing list
>cedet-semantic@...
>https://lists.sourceforge.net/lists/listinfo/cedet-semantic
>
>
>
>
>
>
>
>-------------------------------------------------------
>This SF.net email is sponsored by: ObjectStore.
>If flattening out C++ or Java code to make your application fit in a
>relational database is painful, don't do it! Check out ObjectStore.
>Now part of Progress Software. http://www.objectstore.net/sourceforge
>_______________________________________________
>cedet-semantic mailing list
>cedet-semantic@...
>https://lists.sourceforge.net/lists/listinfo/cedet-semantic
>

>>> "Johan A. Grape" <jeparg@...> seems to think that:
>
>I've been using semantic for a little while, and I have run into a 
>problem with my C++ classes.
>
>As I am sure you all know, when developing for windows, you export 
>classes from dll's by using this construct:
>
>class __declspec(dllexport) classname
>{
>
>
>};
>
>and similarly you import with:
>
>class __declspec(dllimport) classname
>{
>
>
>};
>
>The __declspec() causes semantic to underline the entire contents of the 
>class declaration (which is kind of unreadable), presumably because it 
>looks like garbage to it.
>
>In addition, I develop for cross-platform applications, so I actually 
>have a macro for the import/export, so that depending on the target 
>(.exe or .dll) the macro (on win32) is:
>
>#ifdef BUILD_DLL
>#define EXPORT_DECL __declspec(dllexport)
>#else
>#define EXPORT_DECL __declspec(dllimport)
>#endif
>
>I have taken a poke around in c.bnf, but I cannot see any obvious place 
>to add something to handle the case, and I am a pretty weak lisp 
>programmer....
>
>It seems that what is needed is a generalized mechanism for handling an 
>arbitrary token (defined through some add-to-list?) after the class keyword.
  [ ... ]

For a quick fix in semantic 1.4, you can probably do this:

(add-to-list 'semantic-flex-c-extensions
   '("^\\(EXPORT_DECL\\)$" . semantic-flex-backslash-newline))

and so-on for anything you would like it to just ignore.

the weird function in that list happens to be a function that will
ignore the syntactic match.

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

Does `global-semantic-summary-mode' do what you want?

You may find a long list of useful minor modes in
semantic-util-modes.el

To answer your more specific question, you can use
`semantic-ctxt-current-symbol' to get the text from the buffer which
you can then use to lookup your symbol.

Eric

>>> henrik.jonsson@... seems to think that:
>Hi!
>
>I'm a semantic newcomer. I would like to have a function that displays the
>arguments to the function that I am about to call in the message area. I found
>se-show in the semantic-example.el file and it almost do what I want. If I call
>it it asks for a symbol name, however the default value is already inserted, so
>I only need to press enter to see a beatifull function argument list in the
>message area.
>
>How can I avoid the question of the symbol and only let it read my newly written
>function name?
>It sounds simple, and I have tried to do some changes without success.
>
>(defun se-show (name)
>  "Show important information about some token with NAME.
>Shows how to:
> * Create a variable with `semantic-token->text-functions'.
> * Create the colorized text."
>  (interactive (list (semantic-read-symbol "Symbol: ")))
>  (let ((tok (semantic-find-nonterminal-by-name name (current-buffer) t))
>     parent
>     msg)
>    (if tok
>     (save-excursion
>       ;; Token text functions can take a PARENT argument.  To find this
>       ;; really quickly, jump to TOK, then use the overlay mechanism to
>       ;; Find it.
>       (goto-char (semantic-token-start tok))
>       (setq parent (semantic-current-nonterminal-parent))
>       ;; Create a message
>       (setq msg (funcall se-summary-function tok parent t))
>       ;; Display it
>       (message msg)))))
>
>
>Regards
>
>Henrik
>
>
>
>
>-------------------------------------------------------
>This SF.net email is sponsored by: ObjectStore.
>If flattening out C++ or Java code to make your application fit in a
>relational database is painful, don't do it! Check out ObjectStore.
>Now part of Progress Software. http://www.objectstore.net/sourceforge
>_______________________________________________
>cedet-semantic mailing list
>cedet-semantic@...
>https://lists.sourceforge.net/lists/listinfo/cedet-semantic
>

-- 
          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,

  I think the latest version of ECB has a nifty hack that lets it use
imenu scanned tags within EDB.

  If there is no imenu support for PHP, I suspect you may not be able
to write a BNF grammar for it since PHP is a mix of html and PHP code.
Your best bet may be to write a simple REGEXP based parser.  (see
semantic-texi.el for an example.)

  A better solution is to write an improved lexer that knows how to
"throw out" the useless html and text, leaving only a stream of pure
PHP code.  Then you can write a grammar for it.

  This is best done in the CVS version of semantic which has the great
wisent parser generator (a bison port), and bison->wisent script for
converting old Yacc/C based grammars, and the improved lexer which
would let you follow the above steps.

Have fun
Eric

>>> "Sean O'Connor" <sean@...> seems to think that:
>Hello,
>    I am trying to get ECB (ecb.sourceforge.net) to work with php. From
>what I understand, it depends on semantic for most of the code parsing
>(ie: showing methods). I have looked for a fair amount of time (1-3
>hours) and have not found anything mentioning php.
>
>    Can anyone point me in the right direction for either:
>1) create a simple HelloWorld type syntax/bovine table
>2) adapting existing tables to work with semantic (and ecb)
>3) implementing an existing php syntax/bovine table
>
>    I have these in reverse order for ease of implementation because I
>am intersted in figuring out HOW TO create or adapt existing syntax
>definitions to semantic. The information seems to all be there, but I'm
>a visual (read lazy?) person. 
>
>    I need to start simple, with an easy proof of concept, then build
>out from there. I think I am adept enough to adapt a hello world example
>to a real-world php implementation. 
>
>    At the moment, I don't understand what that involves. I assume it is
>something like:
>- define syntax table
>- convert syntax table to bovine/semantic table (and save to file?)
>- load bovine table in mode (from file?)
>- confirm basic semantic functionality (...?)
>
>Is it feasible to talk about an extremely simple 'foo-mode' which
>perhaps only allows $variables (or something else, syntactically simple)
>and //Comments?
>
>The BNF could then only deal with separating variables from comments. If
>I understand correctly, once the syntax tables are set correctly,
>semantic would then provide a list of known/declared/used variables.
>Perhaps also provide code completion for those variables?
>TIA, 
>
>Sean
>seanoc@...
>



-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
cedet-semantic mailing list
cedet-semantic@...
https://lists.sourceforge.net/lists/listinfo/cedet-semantic

-- 
          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

>>> =?iso-8859-1?q?Ewald=20Ertl?= <eertl@...> seems to think that:
>Hi!
>
>I'm using semanticdb 1.4.4 and Xemacs 21.4 under
>Linux. 
>
>For semantic I've set the following things in my
>.emacs-File: 
>
>;; ###################################################
>;; ##  Semantic - code analyzer                     ##
>;; ###################################################
>(add-to-list 'load-path
>"/usr/share/xemacs/xemacs-packages/lisp/semantic")
>(require 'semantic-load)
>(setq semantic-load-turn-everything-on t)
>(setq semantic-load-turn-useful-things-on t)
> 
>(setq semanticdb-project-roots ( list "/entw/vsop/"  )
>)
>
>(setq semanticdb-default-file-name ".semantic.cache" )
>(global-semanticdb-minor-mode)
> 
>(require 'semantic-ia)
>(global-set-key "\M- " 'semantic-ia-complete-symbol )
> 
>
>In this case when I edit a source-File unter e.g.
>/entw/vsop/src/ and have no Header-File loaded, The
>semantic-ia-completion-symbol does not work. But when
>I have loaded a header-File prior the completion
>work's. 
>
>When setting the 
>(setq semanticdb-default-save-directory
>"~/.emacs_save/semantic")
>
>Even loading a Header-File does not work for me. 
>
>My Question: 
>Is it possible the get semanticdb to load all the
>cache-File's for the project-Directory /entw/vsop? 
>
>If theres also a solution for the
>default-save-directory  would be nice. 
  [ ... ]

If you want certain directories to be loaded at startup, you can put
this in your .emacs file:

(semanticdb-get-database "/path/to/directory")

which will force it to load.  Just make sure all your other variables
and customizations for semantic are set up first.

It does seem useful for a project like area to auto-scan for
databases even if files in them have not been loaded.  I'll add that
to my todo list.

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

Hey,

Jan Sacharuk <jan@...> writes:

> On Tue, 06 May 2003 10:15:46 -0600, Jan Sacharuk <jan@...> wrote:
> 
> So, nobody has any further advice to tell me what I can do to speed
> anything up? I'm personally inclined to blame WinXP a fair amount for
> this, but these performance issues are really quite bad. I don't want
> to turn EVERYTHING off all together, but it's looking more and more
> like I'll have to. Getting a 5 second delay while trying to move
> forward through some code or modify a variable is really entirely
> unacceptable.

Did you try turning off the imenu as I suggested? That's not
EVERYTHING. As I said, I had the same problem which went completely
away when turned off the imenu generation.

-- 
Ole Laursen
http://www.cs.auc.dk/~olau/

That's too bad.  I run with most of the goodies turned on with my
poor little 333Mhz processor, and I don't have this problem.

One thing you can try is "toggle-debug-on-quit", and when you feel a
pause, quick hit C-g to see what it is waiting for.  That, at least,
will give us a clue where to start looking.

Good Luck
Eric

>>> Jan Sacharuk <jan@...> seems to think that:
>On Tue, 06 May 2003 10:15:46 -0600, Jan Sacharuk <jan@...> wrote:
>
>So, nobody has any further advice to tell me what I can do to speed 
>anything up? I'm personally inclined to blame WinXP a fair amount for this, 
>but these performance issues are really quite bad. I don't want to turn 
>EVERYTHING off all together, but it's looking more and more like I'll have 
>to. Getting a 5 second delay while trying to move forward through some code 
>or modify a variable is really entirely unacceptable.
>
>Thanks,
>
>JS
>
>> Incidentally, it didn't help. I still have odd hang-ups when trying to 
>> work. While it's sort of a shame, I'd sort of like to be able to turn off 
>> the little tips that show the syntactic construct that you're working in 
>> (if, while, whatever) or the class or method that you're currently 
>> editing in the command mini-buffer (or whatever it's properly called.)
>

-- 
          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

On Thu, 8 May 2003 23:19:59 -0400, Eric M. Ludlam <eric@...> 
wrote:

> One thing you can try is "toggle-debug-on-quit", and when you feel a
> pause, quick hit C-g to see what it is waiting for.  That, at least,
> will give us a clue where to start looking.

Fantastic. That's exactly the kind of advice I was looking for. I'll try it 
out today. :)

JS

-- 
Jan Sacharuk
Programmer
BioWare Corp.
jan@...

On 08 May 2003 21:47:20 +0200, Ole Laursen <olau@...> wrote:


> Did you try turning off the imenu as I suggested? That's not
> EVERYTHING. As I said, I had the same problem which went completely
> away when turned off the imenu generation.

Oh, yeah. I turned it off a long time ago when the imenu generation was 
costing me five minutes at startup. I have Emacs open all the files I was 
working on when I quit, so the iMenu generation sometimes had to go through 
30 source files. I'm not entirely sure what the imenu stuff does, anyway, 
so I don't really miss it. :)

JS

-- 
Jan Sacharuk
Programmer
BioWare Corp.
jan@...