Home / Browse / CEDET / Mail / Archive

CEDET

Email Archive: cedet-devel (read-only)

Re: [CEDET-devel] CEDET - Semantic parsing problem...

From: David PONCE <david.ponce@wa...> - 2004-06-29 12:25

Hi Javier,

I am sorry for the long delay to reply to your request.

[...]
> I'm seeing a problem with a particular statement in a C-file. In it's
> current form, the methods buffer in ecb does not show any functions that are
> in the c-file. In the acual buffer I see those red underlines everywhere in
> this buffer. From my understanding, those come from
> semantic-show-unmatched-syntax-mode and this in turn causes the methods
> buffer to not show any methods. I have narrowed it down to a particular
> expression.
>  
> Currently there is the #ifdef expression shown immediately below. This sits
> near the top of the file. This is a legal expression in C and the compiler
> has no problems handling it. When I have this in the file, semantic seems to
> now understand the syntax.
> #if ((SOME_POUND_DEFINE1 * SOME_POUND_DEFINE2) > \
>      (SOME_POUND_DEFINE3))
> # error We should never get here!!
> #endif
> 
> If I now modify the #ifdef to look like this, semantic seems to like it and
> the methods now appear in the ecb methods buffer. I do not see any of those
> annoying red underlines. :-)
> #if ((SOME_POUND_DEFINE1 * SOME_POUND_DEFINE2) > (SOME_POUND_DEFINE3))
> # error We should never get here!!
> #endif
> 

I think the problem was in the semantic C lexer, that didn't
correctly handle preprocessor statements.

Please could you try the following patch to
semantic/bovine/semantic-c.el, and tell me if it solves the issue?

Regards.
David

Index: semantic-c.el
===================================================================
RCS file: /cvsroot/cedet/cedet/semantic/bovine/semantic-c.el,v
retrieving revision 1.30
diff -c -r1.30 semantic-c.el
*** semantic-c.el	5 Apr 2004 02:05:43 -0000	1.30
--- semantic-c.el	29 Jun 2004 12:18:58 -0000
***************
*** 49,58 ****
    (setq semantic-lex-end-point (point))
    nil)
  
  (define-lex-regex-analyzer semantic-lex-c-if
    "Ignore various forms of #if/#else/#endif conditionals."
    "^#\\(if\\(def\\)?\\|el\\(if\\|se\\)\\|endif\\)"
!   (when (bolp) (end-of-line))
    (setq semantic-lex-end-point (point))
    nil)
  
--- 49,77 ----
    (setq semantic-lex-end-point (point))
    nil)
  
+ ;;; Compatibility
+ ;;
+ (if (fboundp 'c-end-of-macro)
+     (defalias 'semantic-c-end-of-macro 'c-end-of-macro)
+   ;; From cc-mode 5.30
+   (defun semantic-c-end-of-macro ()
+     "Go to the end of a preprocessor directive.
+ More accurately, move point to the end of the closest following line
+ that doesn't end with a line continuation backslash.
+ 
+ This function does not do any hidden buffer changes."
+     (while (progn
+              (end-of-line)
+              (when (and (eq (char-before) ?\\)
+                         (not (eobp)))
+                (forward-char)
+                t))))
+   )
+ 
  (define-lex-regex-analyzer semantic-lex-c-if
    "Ignore various forms of #if/#else/#endif conditionals."
    "^#\\(if\\(def\\)?\\|el\\(if\\|se\\)\\|endif\\)"
!   (semantic-c-end-of-macro)
    (setq semantic-lex-end-point (point))
    nil)

Thread View

Thread	Author	Date
[CEDET-devel] RE: CEDET - Semantic parsing problem...	Oviedo, Javier <joviedo@te...>

Re: [CEDET-devel] CEDET - Semantic parsing problem...

From: David PONCE <david.ponce@wa...> - 2004-06-29 12:25

Hi Javier,

I am sorry for the long delay to reply to your request.

[...]
> I'm seeing a problem with a particular statement in a C-file. In it's
> current form, the methods buffer in ecb does not show any functions that are
> in the c-file. In the acual buffer I see those red underlines everywhere in
> this buffer. From my understanding, those come from
> semantic-show-unmatched-syntax-mode and this in turn causes the methods
> buffer to not show any methods. I have narrowed it down to a particular
> expression.
>  
> Currently there is the #ifdef expression shown immediately below. This sits
> near the top of the file. This is a legal expression in C and the compiler
> has no problems handling it. When I have this in the file, semantic seems to
> now understand the syntax.
> #if ((SOME_POUND_DEFINE1 * SOME_POUND_DEFINE2) > \
>      (SOME_POUND_DEFINE3))
> # error We should never get here!!
> #endif
> 
> If I now modify the #ifdef to look like this, semantic seems to like it and
> the methods now appear in the ecb methods buffer. I do not see any of those
> annoying red underlines. :-)
> #if ((SOME_POUND_DEFINE1 * SOME_POUND_DEFINE2) > (SOME_POUND_DEFINE3))
> # error We should never get here!!
> #endif
> 

I think the problem was in the semantic C lexer, that didn't
correctly handle preprocessor statements.

Please could you try the following patch to
semantic/bovine/semantic-c.el, and tell me if it solves the issue?

Regards.
David

Index: semantic-c.el
===================================================================
RCS file: /cvsroot/cedet/cedet/semantic/bovine/semantic-c.el,v
retrieving revision 1.30
diff -c -r1.30 semantic-c.el
*** semantic-c.el	5 Apr 2004 02:05:43 -0000	1.30
--- semantic-c.el	29 Jun 2004 12:18:58 -0000
***************
*** 49,58 ****
    (setq semantic-lex-end-point (point))
    nil)
  
  (define-lex-regex-analyzer semantic-lex-c-if
    "Ignore various forms of #if/#else/#endif conditionals."
    "^#\\(if\\(def\\)?\\|el\\(if\\|se\\)\\|endif\\)"
!   (when (bolp) (end-of-line))
    (setq semantic-lex-end-point (point))
    nil)
  
--- 49,77 ----
    (setq semantic-lex-end-point (point))
    nil)
  
+ ;;; Compatibility
+ ;;
+ (if (fboundp 'c-end-of-macro)
+     (defalias 'semantic-c-end-of-macro 'c-end-of-macro)
+   ;; From cc-mode 5.30
+   (defun semantic-c-end-of-macro ()
+     "Go to the end of a preprocessor directive.
+ More accurately, move point to the end of the closest following line
+ that doesn't end with a line continuation backslash.
+ 
+ This function does not do any hidden buffer changes."
+     (while (progn
+              (end-of-line)
+              (when (and (eq (char-before) ?\\)
+                         (not (eobp)))
+                (forward-char)
+                t))))
+   )
+ 
  (define-lex-regex-analyzer semantic-lex-c-if
    "Ignore various forms of #if/#else/#endif conditionals."
    "^#\\(if\\(def\\)?\\|el\\(if\\|se\\)\\|endif\\)"
!   (semantic-c-end-of-macro)
    (setq semantic-lex-end-point (point))
    nil)

RE: [CEDET-devel] CEDET - Semantic parsing problem...

From: Oviedo, Javier <joviedo@te...> - 2004-06-29 13:12

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dus-ascii">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>RE: [CEDET-devel] CEDET - Semantic parsing problem...</TITLE>
</HEAD>
<BODY>

<P><FONT SIZE=3D2>Hello David. This does indeed fix the issue!! Thank =
you.</FONT>
</P>

<P><FONT SIZE=3D2>&nbsp;-------------</FONT>
<BR><FONT SIZE=3D2>&nbsp;Javier</FONT>
<BR><FONT SIZE=3D2>&nbsp;</FONT>
</P>
<BR>
<BR>

<P><FONT SIZE=3D2>-----Original Message-----</FONT>
<BR><FONT SIZE=3D2>From: David PONCE [<A =
HREF=3D"mailto:david.ponce@...>=
] </FONT>
<BR><FONT SIZE=3D2>Sent: Tuesday, June 29, 2004 8:25 AM</FONT>
<BR><FONT SIZE=3D2>To: joviedo@...>
<BR><FONT SIZE=3D2>Cc: cedet-devel@...>
<BR><FONT SIZE=3D2>Subject: Re: [CEDET-devel] CEDET - Semantic parsing =
problem...</FONT>
</P>
<BR>

<P><FONT SIZE=3D2>Hi Javier,</FONT>
</P>

<P><FONT SIZE=3D2>I am sorry for the long delay to reply to your =
request.</FONT>
</P>

<P><FONT SIZE=3D2>[...]</FONT>
<BR><FONT SIZE=3D2>&gt; I'm seeing a problem with a particular =
statement in a C-file. In it's </FONT>
<BR><FONT SIZE=3D2>&gt; current form, the methods buffer in ecb does =
not show any functions </FONT>
<BR><FONT SIZE=3D2>&gt; that are in the c-file. In the acual buffer I =
see those red underlines </FONT>
<BR><FONT SIZE=3D2>&gt; everywhere in this buffer. From my =
understanding, those come from </FONT>
<BR><FONT SIZE=3D2>&gt; semantic-show-unmatched-syntax-mode and this in =
turn causes the </FONT>
<BR><FONT SIZE=3D2>&gt; methods buffer to not show any methods. I have =
narrowed it down to a </FONT>
<BR><FONT SIZE=3D2>&gt; particular expression.</FONT>
<BR><FONT SIZE=3D2>&gt;&nbsp; </FONT>
<BR><FONT SIZE=3D2>&gt; Currently there is the #ifdef expression shown =
immediately below. This </FONT>
<BR><FONT SIZE=3D2>&gt; sits near the top of the file. This is a legal =
expression in C and the </FONT>
<BR><FONT SIZE=3D2>&gt; compiler has no problems handling it. When I =
have this in the file, </FONT>
<BR><FONT SIZE=3D2>&gt; semantic seems to now understand the syntax. =
#if ((SOME_POUND_DEFINE1 </FONT>
<BR><FONT SIZE=3D2>&gt; * SOME_POUND_DEFINE2) &gt; \</FONT>
<BR><FONT SIZE=3D2>&gt;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
(SOME_POUND_DEFINE3))</FONT>
<BR><FONT SIZE=3D2>&gt; # error We should never get here!!</FONT>
<BR><FONT SIZE=3D2>&gt; #endif</FONT>
<BR><FONT SIZE=3D2>&gt; </FONT>
<BR><FONT SIZE=3D2>&gt; If I now modify the #ifdef to look like this, =
semantic seems to like </FONT>
<BR><FONT SIZE=3D2>&gt; it and the methods now appear in the ecb =
methods buffer. I do not see </FONT>
<BR><FONT SIZE=3D2>&gt; any of those annoying red underlines. :-) #if =
((SOME_POUND_DEFINE1 * </FONT>
<BR><FONT SIZE=3D2>&gt; SOME_POUND_DEFINE2) &gt; (SOME_POUND_DEFINE3)) =
# error We should never </FONT>
<BR><FONT SIZE=3D2>&gt; get here!! #endif</FONT>
<BR><FONT SIZE=3D2>&gt; </FONT>
</P>

<P><FONT SIZE=3D2>I think the problem was in the semantic C lexer, that =
didn't correctly handle preprocessor statements.</FONT>
</P>

<P><FONT SIZE=3D2>Please could you try the following patch to =
semantic/bovine/semantic-c.el, and tell me if it solves the =
issue?</FONT>
</P>

<P><FONT SIZE=3D2>Regards.</FONT>
<BR><FONT SIZE=3D2>David</FONT>
</P>

<P><FONT SIZE=3D2>Index: semantic-c.el =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D</FONT>
<BR><FONT SIZE=3D2>RCS file: =
/cvsroot/cedet/cedet/semantic/bovine/semantic-c.el,v</FONT>
<BR><FONT SIZE=3D2>retrieving revision 1.30</FONT>
<BR><FONT SIZE=3D2>diff -c -r1.30 semantic-c.el</FONT>
<BR><FONT SIZE=3D2>*** =
semantic-c.el&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 5 Apr 2004 02:05:43 =
-0000&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 1.30</FONT>
<BR><FONT SIZE=3D2>--- =
semantic-c.el&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 29 Jun 2004 12:18:58 =
-0000</FONT>
<BR><FONT SIZE=3D2>***************</FONT>
<BR><FONT SIZE=3D2>*** 49,58 ****</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; (setq semantic-lex-end-point =
(point))</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; nil)</FONT>
<BR><FONT SIZE=3D2>&nbsp; </FONT>
<BR><FONT SIZE=3D2>&nbsp; (define-lex-regex-analyzer =
semantic-lex-c-if</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; &quot;Ignore various forms of =
#if/#else/#endif conditionals.&quot;</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; =
&quot;^#\\(if\\(def\\)?\\|el\\(if\\|se\\)\\|endif\\)&quot;</FONT>
<BR><FONT SIZE=3D2>!&nbsp;&nbsp; (when (bolp) (end-of-line))</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; (setq semantic-lex-end-point =
(point))</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; nil)</FONT>
<BR><FONT SIZE=3D2>&nbsp; </FONT>
<BR><FONT SIZE=3D2>--- 49,77 ----</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; (setq semantic-lex-end-point =
(point))</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; nil)</FONT>
<BR><FONT SIZE=3D2>&nbsp; </FONT>
<BR><FONT SIZE=3D2>+ ;;; Compatibility</FONT>
<BR><FONT SIZE=3D2>+ ;;</FONT>
<BR><FONT SIZE=3D2>+ (if (fboundp 'c-end-of-macro)</FONT>
<BR><FONT SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp; (defalias =
'semantic-c-end-of-macro 'c-end-of-macro)</FONT>
<BR><FONT SIZE=3D2>+&nbsp;&nbsp; ;; From cc-mode 5.30</FONT>
<BR><FONT SIZE=3D2>+&nbsp;&nbsp; (defun semantic-c-end-of-macro =
()</FONT>
<BR><FONT SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp; &quot;Go to the end of a =
preprocessor directive.</FONT>
<BR><FONT SIZE=3D2>+ More accurately, move point to the end of the =
closest following line </FONT>
<BR><FONT SIZE=3D2>+ that doesn't end with a line continuation =
backslash.</FONT>
<BR><FONT SIZE=3D2>+ </FONT>
<BR><FONT SIZE=3D2>+ This function does not do any hidden buffer =
changes.&quot;</FONT>
<BR><FONT SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp; (while (progn</FONT>
<BR><FONT =
SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp;&nbsp; (end-of-line)</FONT>
<BR><FONT =
SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp;&nbsp; (when (and (eq (char-before) ?\\)</FONT>
<BR><FONT =
SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp; (not (eobp)))</FONT>
<BR><FONT =
SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp;&nbsp;&nbsp;&nbsp; (forward-char)</FONT>
<BR><FONT =
SIZE=3D2>+&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp;&nbsp;&nbsp;&nbsp; t))))</FONT>
<BR><FONT SIZE=3D2>+&nbsp;&nbsp; )</FONT>
<BR><FONT SIZE=3D2>+ </FONT>
<BR><FONT SIZE=3D2>&nbsp; (define-lex-regex-analyzer =
semantic-lex-c-if</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; &quot;Ignore various forms of =
#if/#else/#endif conditionals.&quot;</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; =
&quot;^#\\(if\\(def\\)?\\|el\\(if\\|se\\)\\|endif\\)&quot;</FONT>
<BR><FONT SIZE=3D2>!&nbsp;&nbsp; (semantic-c-end-of-macro)</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; (setq semantic-lex-end-point =
(point))</FONT>
<BR><FONT SIZE=3D2>&nbsp;&nbsp;&nbsp; nil)</FONT>
<BR><FONT SIZE=3D2>&nbsp; </FONT>
</P>

</BODY>
</HTML>

Re: [CEDET-devel] CEDET - Semantic parsing problem...

From: David PONCE <david.ponce@wa...> - 2004-06-29 13:33

> Hello David. This does indeed fix the issue!! Thank you.

Great!  So, I will commit the fix.

Thanks.
David

[CEDET-devel] RE: CEDET - Semantic parsing problem...

From: Eric M. Ludlam <eric@si...> - 2004-07-12 17:47

>Hello all. Does anyone have any thoughts on the mail I sent out this past
> Friday? Please see below. Thanks!
>  
>  
> 
>  ------------- 
>  Javier 

Hi Javier,

  I'm finally going back through the archives of email I missed, and
found all your mail starting around June 17th.  I will try to
summarize what I know from your various messages.  I don't know which
of your troubles you have already solved.  I can't respond to you
directly because the mailing list archive won't display email
addresses.  In addition, the archive seems to have deleted the body of
some of your messages.

  WinXP build problem:

I don't have WinXP to try a build on.  On the JDEE mailing list,
folks who had trouble used the cygnus build environment.  I couldn't
find a link to the specific article.  Perhaps someone else remembers.
Since you asked more questions later, I'll assume you got past this.

  Parse problem involving #ifdef:

The red underlines are part of `semantic-show-unmatched-syntax-mode'.
The examples involved some #ifdef calls using \ line endings for
continuation.  The pre-processor symbol handling isn't very good but
it doues identify the \ and the end of the line.  In my experiment
(just now) I could replicate the problem you describe if I put a space
after the line ending \.  This is related to the function
c-end-of-macro so that is probably better fixed in cc-engine.el

  Echo area issue:

Semantic does have several modes that display stuff in the echo area.
However, I cannot reproduce, nor do I recognize the specific message
that you put in you note.  I could find no message statement that
might produce such a text string so I suspect it is some sort of
error.

I think David recently added in some fixes to "put back" whatever was
in the minibuffer but that won't work too well if it is an error
instead.  If it is an uncaught error, then it should be possible to
get a stack trace with "M-x toggle-debug-on-error".

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