CEDET

Michal <rabbitko@...> writes:
> I applied ede on simple project.
> Whole project is short and in single directory, without subdirectories:
>
> Human.cpp
> Human.h
> example.cpp
>
> code is simple.

[...]

> Now when I point with cursor within "present" in line [1], then M-x
> semantic-ia-fast-jump jumps to Human::present declaration ([2]).
> Then I use M-x semantic-analyze-proto-impl-toggle to get to definition,
> but this fails with a message:
>
> Could not find suitable implementation

Yes, that feature is a bit tricky.

If Semantic has never seen the file Human.cpp, it will indeed not find
the implementation of Human.cpp. It should work if you load Human.cpp
and give Semantic the time to parse it (call 'M-x bovinate' to enforce
it).

You can get a dump of the current database by calling

M-x semanticdb-dump-all-table-summary

Hit enter on the DBs containing your project files to open it. Then hit
enter on the :tables entry to open the list of files. If you see
something like

*#<semanticdb-project-database-file human/ (3 tables DIRTY)>
   ] Name: "human/"
   ] Class: #'semanticdb-project-database-file
   ] reference-directory "/data/void/human/"
   ] new-table-class #'semanticdb-table
   ] cache : nil
   ] :tables #<list o' stuff: 3 entries>
     > #<semanticdb-table example.cpp (2 tags), DIRTY>
     > #<semanticdb-table Human.h (1 tags)>
     > #<semanticdb-table Human.cpp (2 tags), DIRTY>


it means that is has parsed two tags in Human.cpp, and it then should be
able to find the implementation. This should also work between sessions,
since this information gets saved to disk.

> My concern is that when I am in Human.h buffer, ede-object value is nil,
> which (probably) means that this file does not belong to the project and
> maybe this is a reason.

You can easily add Human.h to the project (in fact, EDE should ask you
as soon as you open it). It will then get added to 'auxsource'.

Regards,
David

Another tid-bit of information - it seems like Semantic is firing on idle
and I'm getting this error in the debug frame:

Idle Work Including Error: "#<buffer .emacs> - Wrong type argument:
number-or-marker-p, (512 39 . 8739)"

What does it mean?

Farhan

On Thu, Aug 27, 2009 at 10:39 AM, Farhan Ahmed <ahmedf@...> wrote:

> Hi there,
>
> I'm following instructions on
> http://cedet.sourceforge.net/intellisense.shtml to enable auto-completion
> of code in Emacs 23.1.
>
> Here's the relevant part from my .emacs file:
>
> (load-file "D:/tools/cedet-1.0pre6/common/cedet.el")
> (global-ede-mode 1)                      ; Enable the Project management
> system
> (semantic-load-enable-code-helpers)      ; Enable prototype help and smart
> completion
> (require 'semantic-ia)
> I don't see any errors while loading Emacs and I've started it in
> -debug-init mode as well.
>
> Here's some sample code I've written to test out the auto-completion:
>
> #include <stdio.h>
> class A {
>   public:
>  A();
>  ~A();
>  int x;
>  int y;
>  int z;
>  int Init();
>   private:
>  int DoIt();
> };
>
> int main(int argc, char *argv[]) {
>  A* ca = new A();
>  ca->
> }
> When I put the cursor at the end of ca->, and then type M-x
> semantic-ia-complete-symbol-menu, I don't get any options for
> auto-completion. I only see "Cannot find types for "ca"" in the mini-buffer.
>
> In the debug window, I see:
>
> let*: Cannot find types for `"ca"'
> Cannot find types for `"ca"' [2 times]
> What am I doing wrong? Do I need to enable some option for Semantic to
> analyze my code or does it do it on demand when I try to use an
> auto-completion option? Also, how do I let SemanticDB analyze and cache
> pre-existing 'include' files so that I can get auto-completion for those
> functions?
>
> Any help would be greatly appreciated!
>
> Farhan
>
>

Farhan Ahmed <ahmedf@...> writes:
> I'm following instructions on
> http://cedet.sourceforge.net/intellisense.shtml to enable auto-completion of
> code in Emacs 23.1.
>
> Here's the relevant part from my .emacs file:
>
> (load-file "D:/tools/cedet-1.0pre6/common/cedet.el")

Are you using CEDET on Windows? If so, I'm actually surprised you got
1.0pre6 compiled in the first place, since there were some bugs
regarding the Windows platform in that version.

Please update to the CVS version, which you can get by doing

cvs -z3 -d:pserver:anonymous@... co -P cedet

> When I put the cursor at the end of ca->, and then type M-x
> semantic-ia-complete-symbol-menu, I don't get any options for
> auto-completion. I only see "Cannot find types for "ca"" in the mini-buffer.

Works for me. Put the cursor behind the 'ca->' and do

M-x semantic-analyze-debug-assist

to get further output on why the completion does not work.

> Another tid-bit of information - it seems like Semantic is firing on
> idle and I'm getting this error in the debug frame:
> Idle Work Including Error: "#<buffer .emacs> - Wrong type argument:
> number-or-marker-p, (512 39 . 8739)"

It means that Semantic had a problem parsing some stuff in your .emacs
file, which you probably had opened in your Emacs session. This doesn't
interfere with the C++ parsing, so that error shouldn't matter here at
the moment.

Regards,
David

Yes, Piotr also suggested using the CVS version. I did - and it all works
now! And yes, I am using it on Windows. I didn't pay attention during the
compilation, so perhaps things did break.

Appreciate all the help guys.

Farhan

On Thu, Aug 27, 2009 at 12:57 PM, David Engster <deng@...:

> Farhan Ahmed <ahmedf@...> writes:
> > I'm following instructions on
> > http://cedet.sourceforge.net/intellisense.shtml to enable
> auto-completion of
> > code in Emacs 23.1.
> >
> > Here's the relevant part from my .emacs file:
> >
> > (load-file "D:/tools/cedet-1.0pre6/common/cedet.el")
>
> Are you using CEDET on Windows? If so, I'm actually surprised you got
> 1.0pre6 compiled in the first place, since there were some bugs
> regarding the Windows platform in that version.
>
> Please update to the CVS version, which you can get by doing
>
> cvs -z3 -d:pserver:anonymous@... co
> -P cedet
>
> > When I put the cursor at the end of ca->, and then type M-x
> > semantic-ia-complete-symbol-menu, I don't get any options for
> > auto-completion. I only see "Cannot find types for "ca"" in the
> mini-buffer.
>
> Works for me. Put the cursor behind the 'ca->' and do
>
> M-x semantic-analyze-debug-assist
>
> to get further output on why the completion does not work.
>
> > Another tid-bit of information - it seems like Semantic is firing on
> > idle and I'm getting this error in the debug frame:
> > Idle Work Including Error: "#<buffer .emacs> - Wrong type argument:
> > number-or-marker-p, (512 39 . 8739)"
>
> It means that Semantic had a problem parsing some stuff in your .emacs
> file, which you probably had opened in your Emacs session. This doesn't
> interfere with the C++ parsing, so that error shouldn't matter here at
> the moment.
>
> Regards,
> David
>
>
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
> trial. Simplify your report design, integration and deployment - and focus
> on
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> cedet-semantic mailing list
> cedet-semantic@...
> https://lists.sourceforge.net/lists/listinfo/cedet-semantic
>

Pieter Swinkels <swinkels.pieter@...> writes:
> A simple question here about semantic and the STL header files. The
> STL header files do not have an extension, they are simply called
> "vector" or "algorithm" instead of "vector.h" and "algorithm.h". How
> can I let semantic know that it should parse such a file as a C++
> file?

If you system include paths are setup correctly, this should work right
away.

For testing this, create a test C++ file, for example

------------------
#include <vector>
int main(void)
{
     std::vector<int> v;
     v. // Put cursor behind dot and run M-x semantic-analyze-debug-assist
}
-----------------

Do as the comment says, and it should tell you all kinds of stuff in a
new buffer. If you see there something like

Semantic has found the datatype class vector {}
and its list of members.

then everything should work just fine. If it says that it could not find
the type, then take a look further down at the system include path. It
should include the paths which have the STL header files, for example

  /usr/include/c++/4.3/

If it does not contain this path, and you're using gcc, Semantic somehow
failed to parse the includes from your gcc installation. If you're not
using gcc, you have to add the system include paths manually, for
example by using semantic-add-system-include or by creating a project
using EDE.

> I can imagine this question has been asked before, but searching in the
> mailing list archive seems not possible.

You can search the CEDET/Semantic groups at Gmane:

http://search.gmane.org/form.php?group=gmane.emacs.cedet
http://search.gmane.org/form.php?group=gmane.emacs.semantic

Regards,
-David