Hi Durk,
while trying to analyze the crash I got to the conviction, that the call
stack is wrong and points not to the "right" position (there are some
inconsistencies in the call stack). I think this is due to the missing
debug information in the system libraries. I try to get the error with
the debug version over night.
Maik
Durk Talsma schrieb am 16.11.2006 22:16:
> Hi Maik,
>
> It looks like I completely overlooked your message yesterday. I've been
> playing with your idea a bit. Some more comments below.
>
> On Wednesday 15 November 2006 19:53, Maik Justus wrote:
>
>> Hi,
>>
>> I built a version with debug info only for this file and got, as
>> expected, normal performance.
>> The problem is in line 807:
>> if (i != current)
>> I have not much experience with overloaded comparators, but it seems to
>> me, that not the "pointers" are compared in this case, but the complete
>>
>
> That is not what is supposed to happen: I'm not comparing whether the two
> objects the iterators are pointing to are identical or not, but whether the
> two iterators themselves are pointing to the same object.
> 'current' is pointing to a valid object (which is established at the
> beginning
> of the function, and "i" is also pointing to a valid object (because it
> iterates between towerController->getActiveTraffic().begin(); and
> towerController->getActiveTraffic().end()
>
> the overloaded object comparison would be something like:
>
> if ((*current) == (*i))
>
> I've added a bool operator == (const FGTrafficRecord &right) const {return id
> == right.id;} as you suggested, and added a few debug warning messages to it
> to see whether it got called, which it didn't. So, I'm pretty sure that the
> code above is actually just comparing whether the two iterators are pointing
> to the same object or not.
>
>
>> structure. But "current" holds an invalid pointer in its structure:
>> (((*(current)._Myptr).runway))._Bx
>>
>
> I did find out that I never assigned a value to runway, in case of traffic
> that is on the ground network (whereas I do assign a runway for traffic that
> is waiting for take-off).
>
>
>> I don't know how this can happen, that the string "runway" contains a
>> invalid pointer, but this seems to cause the crash. Maybe another
>> pointer writing on this pointer? Or did you copy the address of this
>> FGTrafficRecord (or of the string runway) and call the destructor?
>>
>
> Hmm, no, not likely: I'm not acessing the runway string other than by copying
> a value into it. So assuming stl string has a correct copy constructor, this
> should work out okay.
>
>
>> For performance reason I would suggest to add an operator == to class
>> FGTrafficRecord, esp.:
>> bool operator == (const FGTrafficRecord &right) const {return id ==
>> right.id;}
>> if "id" is sufficient to compare two FGTrafficRecords.
>>
>>
> Yes, that would be sufficient, is you were comparing objects, but here is
> really is only comparing pointers.
>
> Still, the code still seems to crash on your system. so something's gotta be
> wrong.
>
>
> Cheers,
> Durk
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys - and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Flightgear-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/flightgear-devel
>
>
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Flightgear-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/flightgear-devel
