CEDET

Hi,

On Mon, Dec 27, 2010 at 8:38 PM, Daniel Lopez <dlopez@...> wrote:
> When I'm calling an overloaded function, ie. which has more than one
> prototype, the "Show Tag Summaries" feature shows one of the prototypes
> (maybe the first one found) in the minibuffer as I start typing the
> arguments. Can I cycle through the other available prototypes?
>

Have you tried `semantic-ia-show-variants'?

> Secondly, from searching the archives here it sounds like I might be able to
> get some better features by installing the latest CEDET from version
> control. Are the Bazaar and CVS repositories identical? I assumed the Bazaar
> one was newer, and installed Bazaar which I've never used before, and can't
> work out how to make it download the files. I've looked at various
> documents, to no avail - what's the equivalent to eg. 'cvs checkout ...' or
> 'git clone ...'?
>

The directory structure for the cedet bzr repo is a little messed up and
the instruction on sourceforge doesn't reflect that. Do this to check
out the HEAD.

$ bzr branch bzr://cedet.bzr.sourceforge.net/bzrroot/cedet/code/trunk cedet

To update later, you can do

$ bzr merge --pull

> Daniel
>

Hope this was helpful.

-- 
Suvayu

Open source is the future. It sets us free.

On 28/12/2010 05:12, suvayu ali wrote:
> The directory structure for the cedet bzr repo is a little messed up
> and the instruction on sourceforge doesn't reflect that. Do this to
> check out the HEAD.
>
> $ bzr branch
> bzr://cedet.bzr.sourceforge.net/bzrroot/cedet/code/trunk cedet
>
> To update later, you can do
>
> $ bzr merge --pull
>

Thanks, that worked a treat.


> Have you tried `semantic-ia-show-variants'?
>

Having partial success with this.

If I type eg. "printf", then do M-x semantic-ia-show-variants, it shows me signatures for printf and printf_s - this is OK. (These aren't actually overloaded functions, but as I'm having some other problems I'll come back to that later.)

However if I add the opening parenthesis, ie. "printf(", then do M-x semantic-ia-show-variants, then I get the error, "Wrong type argument: integer-or-marker-p, nil".

Single-stepping through the Lisp:

> (defun semantic-ia-show-variants (point)
>   "Display a list of all variants for the symbol under POINT."
>   (interactive "P")
>   (let* ((ctxt (semantic-analyze-current-context point))
> 	 (comp nil)
> 	 )
>
>     ;; We really want to look at the function if we are on an
>     ;; argument.  Are there some additional rules we care about for
>     ;; changing the CTXT we look at?
>     (when (semantic-analyze-context-functionarg-p ctxt)
>       (goto-char (cdr (oref ctxt bounds)))
>       (setq ctxt (semantic-analyze-current-context (point))))

"(oref ctxt bounds)" returns 'nil' and then cdr raises the error. (The 'when' block is skipped in the first instance when I don't type an opening parenthesis, avoiding the error).

Don't know the object-oriented parts of Lisp yet so I've got to pause here.

By the way, this is how I'm setting up CEDET in my init file, in case it's important:

> ;;; CEDET
>
> ;; Ignore CEDET that's bundled with Emacs
> (setq load-path (remove "d:/Programs/Text/emacs-23.2/lisp/cedet" load-path))
>
> ;; Load CEDET from BZR which is within site-lisp directory
> (load "cedet.el")
>
> ;;
> (semantic-add-system-include "D:/Prog/MS/Microsoft Visual Studio 9.0/VC/include" 'c-mode)
> (semantic-add-system-include "D:/Prog/MS/Microsoft Visual Studio 9.0/VC/include" 'c++-mode)
> (semantic-add-system-include "D:/Prog/MS/Microsoft Platform SDK for Windows Server 2003 R2/Include" 'c-mode)
> (semantic-add-system-include "D:/Prog/MS/Microsoft Platform SDK for Windows Server 2003 R2/Include" 'c++-mode)
> (semantic-add-system-include "D:/Prog/C/boost/boost_1_37" 'c-mode)
> (semantic-add-system-include "D:/Prog/C/boost/boost_1_37" 'c++-mode)
> (semantic-add-system-include "D:/Prog/dan/C/reusable/checkout" 'c-mode)
> (semantic-add-system-include "D:/Prog/dan/C/reusable/checkout" 'c++-mode)
>
> (semantic-load-enable-minimum-features)
>
> ;; This enables parsing of header files.
> (setq semantic-idle-work-update-headers-flag t)
>
> ;; Tag decoration - yes: includes, no: tag boundaries
> (global-semantic-decoration-mode 1)
> (require 'semantic-decorate-include)
> (setq semantic-decoration-styles '(("semantic-decoration-on-includes" . t) ("semantic-decoration-on-protected-members") ("semantic-decoration-on-private-members")))
>
> (global-semantic-mru-bookmark-mode 1)
>
> (global-senator-minor-mode 1)


Daniel

Hi Daniel,

On Tue, Dec 28, 2010 at 2:44 PM, Daniel Lopez <dlopez@...> wrote:
>
>> Have you tried `semantic-ia-show-variants'?
>>
>
> Having partial success with this.
>
> If I type eg. "printf", then do M-x semantic-ia-show-variants, it
> shows me signatures for printf and printf_s - this is OK. (These
> aren't actually overloaded functions, but as I'm having some other
> problems I'll come back to that later.)
>
> However if I add the opening parenthesis, ie. "printf(", then do M-x
> semantic-ia-show-variants, then I get the error, "Wrong type argument:
> integer-or-marker-p, nil".
>
> Single-stepping through the Lisp:

I am not sure what is the problem with your setup. I am a lisp
novice. But I have been using CEDET for quite sometime now with a C++
data analysis framework on linux.

This is what I get when I try `semantic-ia-show-variants' with my
framework:

> if (integral) h_sum->Add(h_sample[i]); // cursor is on the last d of Add
...
> virtual void AddBinContent (Int_t bin,Double_t w)
> virtual void AddBinContent (Int_t bin)
> virtual void Add (const TH1* h,const TH1* h2,Double_t c1,Double_t c2)
> virtual void Add (const TH1* h1,Double_t c1)
> virtual void Add (TF1* h1,Double_t c1,Option_t* option)
> void AddAt (Float_t c,Int_t i)
> -UUU:%%--F1  *Symbol Variants*   All (1,0)      (Help View)

As for your setup, it is fairly similar with mine. This is what I do to
setup CEDET:

https://github.com/suvayu/.emacs.d/blob/master/lisp/cedet-configs.el#L70

(lines 70-78, 132-133)

And then I call (semantic-default-c-setup).

I hope that helps.

--
Suvayu

Open source is the future. It sets us free.

David Engster writes:
> Yet another C++ namespace issue... The library 'Armadillo' uses
> constructs like these:
>
> namespace foo {
>   #include "bar/baz.hpp"
> }

Here's a suggestion. First, parse those includes with a special
attribute 'inside-ns':

--- semantic/bovine/c.by        2010-10-07 19:46:02 +0000
+++ semantic/bovine/c.by        2010-12-14 12:43:55 +0000
@@ -376,6 +376,9 @@
  ;; PUBLIC or PRIVATE bits.  Ignore them for now.
   | template
   | using
+ ;; Includes inside namespaces
+  | spp-include
+    (TAG $1 'include :inside-ns t)
   | ;;EMPTY
   ;

Then we have to make sure Semantic finds those includes inside a
namespaces...

(defun semantic-find-tags-included (&optional table)
  "Find all tags in TABLE that are of the 'include class.
TABLE is a tag table.  See `semantic-something-to-tag-table'."
  (let ((tags (semantic-find-tags-by-class 'include table))
	(namespaces (semantic-find-tags-by-type "namespace" table)))
    (dolist (cur namespaces)
      (setq tags
	    (append tags
		    (semantic-find-tags-by-class 'include (semantic-tag-get-attribute cur :members)))))
    tags))

...and use a new override to wrap the tags in those includes inside a
namespace type:

(define-mode-local-override semanticdb-find-table-for-include c-mode
  (includetag &optional table)
  ""
  (let ((inctable (semanticdb-find-table-for-include-default includetag table))
	tags newtags namespace prefix)
    (when (and (semantic-tag-get-attribute includetag :inside-ns)
	       table)
      ;; Find namespace which had this include
      (setq namespaces (semantic-find-tags-by-type "namespace" table))
      (when (and namespaces
		 (slot-boundp inctable 'tags))
	(dolist (cur namespaces)
	  (when (semantic-find-tags-by-name (semantic-tag-name includetag) (semantic-tag-get-attribute cur :members))
	    (setq prefix (semantic-tag-name cur))))
	(unless (semantic-find-tags-by-name
		 prefix
		 (semantic-find-tags-by-type "namespace" inctable))
	  (setq tags (oref inctable tags))
	  (setq newtags (list (semantic-tag-new-type prefix "namespace" tags nil)))
	  (oset inctable tags newtags))))
    inctable))


It's just a suggestion. I doesn't work recursively yet, but before
digging further I'd like to know if the above makes sense. I'm not
terribly happy with the extension of `semantic-find-tags-included',
since I guess this is more or less C++-specific.

-David

David Engster writes:
> It's just a suggestion. I doesn't work recursively yet, but before
> digging further I'd like to know if the above makes sense. I'm not
> terribly happy with the extension of `semantic-find-tags-included',
> since I guess this is more or less C++-specific.

Turns out I only had to make a few little changes to make this stuff
correctly parsing the 'Armadillo' library, which is pretty neat. I've
attached the patch I'm currently using.

-David

Am 04.12.2010 um 15:27 schrieb Juhyung Lee:

> The semantic demand for  the objective-C is growing.  Is there any ongoing work for it or anyone who already have done it?
> I'm trying to make it work, but it isn't easy for me now.


Yes, I would love to see that, too!

Best regards,

Arne Schmitz

-- 
Dipl.-Inform. Arne Schmitz              Phone   +49 (0)241 80-21817
Computer Graphics Group                 Mobile  +49 (0)151 29145947
RWTH Aachen University                  Fax     +49 (0)241 80-22899
Ahornstrasse 55, 52074 Aachen, Germany  http://www.rwth-graphics.de