CEDET

On 09/19/2012 08:25 AM, Ash Shevlyakov wrote:
> Greetings!
>
> I'm having trouble using code completion with CEDET.
>
> The following error is produced when auto-completion is invoked:
>
> Idle Service Error semantic-idle-summary-idle-function: "#<buffer
> main.cpp>  - Wrong type argument: stringp, (((0) \"GLchar\"))" [2
> times]
>
> main.cpp code:
>
> #include "GL/glext.h"
>
> int main(int argc, char* argv[])
> {
>    return 0;
> }
>
> I'm attaching the result of a backtrace, but it contains lots of
> strange symbols.
> In the absence of #include "GL/glext.h" completion behaves normally.
> What could cause this effect?

Hi,

I was not able to reproduce the problem on my Linux box using the latest 
CEDET from bzr.

The error is reminiscent of problems in older versions of CEDET dealing 
with preprocessor macros.   In your version of glext.h, what is the 
definition of GLchar?  does it use macro expansion, or anything tricky 
like that?  If so, you might need to upgrade your CEDET, or if you are 
already using CEDET 1.1 or later, knowing how the macro is defined in 
glext.h on your system would be a help for debugging.

Eric

On Tue, Sep 25, 2012 at 4:23 AM, Eric M. Ludlam <ericludlam@...> wrote:
> On 09/19/2012 08:25 AM, Ash Shevlyakov wrote:
>>
>> Greetings!
>>
>> I'm having trouble using code completion with CEDET.
>>
>> The following error is produced when auto-completion is invoked:
>>
>> Idle Service Error semantic-idle-summary-idle-function: "#<buffer
>> main.cpp>  - Wrong type argument: stringp, (((0) \"GLchar\"))" [2
>> times]
>>
>> main.cpp code:
>>
>> #include "GL/glext.h"
>>
>> int main(int argc, char* argv[])
>> {
>>    return 0;
>> }
>>
>> I'm attaching the result of a backtrace, but it contains lots of
>> strange symbols.
>> In the absence of #include "GL/glext.h" completion behaves normally.
>> What could cause this effect?
>
>
> Hi,
>
> I was not able to reproduce the problem on my Linux box using the latest
> CEDET from bzr.
>
> The error is reminiscent of problems in older versions of CEDET dealing with
> preprocessor macros.   In your version of glext.h, what is the definition of
> GLchar?  does it use macro expansion, or anything tricky like that?  If so,
> you might need to upgrade your CEDET, or if you are already using CEDET 1.1
> or later, knowing how the macro is defined in glext.h on your system would
> be a help for debugging.
>
> Eric

I'm using revision 8333 of CEDET from bzr, EMACS 24.1.1.

The declaration of GLchar from glext.h is

#ifndef GL_VERSION_2_0
/* GL type for program/shader text */
typedef char GLchar;
#endif

Are any additional steps necessary due to #ifndef?

On 09/25/2012 04:51 AM, Ash Shevlyakov wrote:
> I'm using revision 8333 of CEDET from bzr, EMACS 24.1.1.
>
> The declaration of GLchar from glext.h is
>
> #ifndef GL_VERSION_2_0
> /* GL type for program/shader text */
> typedef char GLchar;
> #endif
>
> Are any additional steps necessary due to #ifndef?

Hmmm, that looks like the same header file I have.   You might simply 
need to delete the contents of your .semanticdb directory, or the file 
associated with /usr/include/GL and let it start over with those tables.

Eric

On Tue, Sep 25, 2012 at 4:29 PM, Eric M. Ludlam <ericludlam@...> wrote:
> On 09/25/2012 04:51 AM, Ash Shevlyakov wrote:
>>
>> I'm using revision 8333 of CEDET from bzr, EMACS 24.1.1.
>>
>> The declaration of GLchar from glext.h is
>>
>> #ifndef GL_VERSION_2_0
>> /* GL type for program/shader text */
>> typedef char GLchar;
>> #endif
>>
>> Are any additional steps necessary due to #ifndef?
>
>
> Hmmm, that looks like the same header file I have.   You might simply need
> to delete the contents of your .semanticdb directory, or the file associated
> with /usr/include/GL and let it start over with those tables.
>
> Eric

Indeed, this seems to have fixed it. Thanks a lot!