gdb error: Previous frame inner to this frame (corrupt stack?)

[Sven]

Hello,

I have a crashing application and tried to find out what is going
wrong. So I run this application with the gdb debugger. I found the
position of the crash in a function which searches the id of a
database element by its name. If the database element is not found it
shall throw an exception. This is the related code:

SRC_ID_TYP Source::getId( const std::string& name_ )
throw( XmlDatabaseExce ption ) {
SRC_ID_TYP srcId;
if ( GetSourceId( name_.c_str(), &( srcId ) ) != OLD_PAG ) {
string errMsg( "Source '" );
errMsg += name_ + "' does not exist in the database";
throw XmlDatabaseExce ption( errMsg );
}
return srcId;
} // *** Source::getId() ***

If the id can be resolved correctly the application works fine. But if
not, it crashes when it tries to throw the exception. The message I
get in the debugger then is:
"Previous frame inner to this frame (corrupt stack?)"

But when I look at the backtrace of the stack it looks very ok:

8 xmldb::Source:: getId() Source.cpp:398 0x08062e68
7 xmldb::Source:: create() Source.cpp:106 0x0805fb3f
6 xmldb::ADSFile: :parseDOMTree() ADSFile.cpp:380 0x0805aec0
5 xmldb::ADSFile: :parseXML() ADSFile.cpp:312 0x0805a0cb
4 xmldb::ADSFile: :read() ADSFile.cpp:104 0x08056fcf
3 ImportExport::i mportData() ImportExport.cp p:373 0x08051c1b
2 ImportExport::e xecute() ImportExport.cp p:305 0x08051764
1 main() iet.cpp:56 0x08055fbb

Know I'm at the end of my knowledge.
Can someone tell me, what is the reason for such an error message? And
what can I investigate further to find the mistake?

Thanks,
Sven

[Sven]

I found the mistake!

With the usage of a factory method I had to move the search of the id
into this static method. But now the throw() declarations of one
method above does not fit any more.

But I wonder, why this is not caught by the compiler. I did not think
that a wrong throw() declaration of a method causes a runtime error.
Isn't it just for code clearance or use the compiler this information
for optimizing?

Greetz, Sven

[tragomaskhalos]

On 30 Nov, 11:19, Sven <[e-mail].d ewrote:

I found the mistake!

With the usage of a factory method I had to move the search of the id
into this static method. But now the throw() declarations of one
method above does not fit any more.

But I wonder, why this is not caught by the compiler. I did not think
that a wrong throw() declaration of a method causes a runtime error.
Isn't it just for code clearance or use the compiler this information
for optimizing?

And the moral of the story is ...
don't use exception specifications.