Re: [Boost-users] [multi_index] SIGABRT in multi_index_container constructor

From: Manuel Jung (gzahl_at_[hidden])
Date: 2007-02-27 02:41:18

Am 26.02.2007, 23:32 Uhr, schrieb Joaquin M Lopez Munoz <joaquin_at_[hidden]>:

> Manuel Jung <gzahl <at> arcor.de> writes:
>
>>
>> Hi,
>>
>> Sorry for hte delay, but i was in a train again yesterday, with no
>> internet...
>> But here im back now.
>>
>> > I don't really know :) I'm trying to gain more info on the
>> > context where the core happens. Let's begin with say the
>> > header and then we will pull from this thread if necessary,
>> > OK?
>>
>> you got two files with this message. These are the slots.h and  
>> octopus.h.
>> They are the only ones with multi_index_container definitions. Tell me  
>> what
>> you need else. The files are just how they are, nothing is stripped out.
>
> Hello, thanks for the files. There's a detail that smells
> funny: octopus.h begins with
>
> #if !defined(NDEBUG)
> #define BOOST_MULTI_INDEX_ENABLE_INVARIANT_CHECKING
> #define BOOST_MULTI_INDEX_ENABLE_SAFE_MODE
> #endif
>
> (which is commented out in the version you've sent, but
> I guess it is when in effect that the crash happens, right?)
Yeah, sure. Just commented out, so i can do my work, without minding my  
little problem.

> whereas slot.h does not have these macros, unless they're
> defined in diver.h or in myconnection.h.
No, they dont have them.

> Could it be that Boost.MultiIndex headers are included by
> diver.h or myconnection.h or slot.cpp (assuming this is
> the name where queue::AddQuery etc. are define) *without*
> defining safe mode and invariant checking macros?
Yeah, this is true. The headers where defined in the diver.h, because i  
was trying and moving code a little bit around at the beginning of using  
Boost.MultiIndex.
I have now fixed it. I dont get an error yet, but a i cannot run a full  
test, because some of the functions are broken at the moment. Ill fix that  
this evening and than well see whats up.

> If so, you've got a problem: these macros have to be
> defined/undefined *globally*. i.e. you cannot compile one
> .cpp with them turned on and other without them: object
> incompatibility ensues.

Is this in the documentation and i didn realize? Maybe it should be said  
explicit, if its not in there.

> Looking forward to your feedback,

Some more results in about 12 hours.
Greetings
Manuel Jung

> Joaquín M López Muñoz
> Telefónica, Investigación y Desarrollo
>
> PS: I answer your other question at a separate thread.
>
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users

-- 
Erstellt mit Operas revolutionärem E-Mail-Modul: http://www.opera.com/mail/