Re: Variable declaration: Global vs. Function
From: Thomas Matthews (Thomas_MatthewsSpamBotsSuck_at_sbcglobal.net)
Date: 01/26/05
- Next message: Aslan Kral: "Re: Good use for Unions"
- Previous message: Andre Kostur: "Re: Good use for Unions"
- In reply to: Mike Wahler: "Re: Variable declaration: Global vs. Function"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 26 Jan 2005 19:45:04 GMT
Mike Wahler wrote:
> "Thomas Matthews" <Thomas_MatthewsSpamBotsSuck@sbcglobal.net> wrote in
> message news:41F7B848.5020906@sbcglobal.net...
>
>>Hi,
>>
>>I'm getting linking errors when I declare a variable in the
>>global scope, but not inside a function. The declarations
>>are the same (only the names have been changed...).
>>
>>class Book
>>{
>> public:
>> Book()
>> { ; }
>> virtual ~Book() { ; }
>
>
> Note that those empty statements in your
> ctor and dtor are not needed.
Noted, but I do that for sytlistic purposes.
So, what's a little wasted compiler time, in
the big picture? {Rhetorical}
>
>
>> string get_title(void) const
>> { return title;}
>> private:
>> std::string title;
>> std::string author;
>>};
>>
>>/* Global scope declaration */
>>Book Global_Book;
>>
>>/* Inside function declaration */
>>void Any_Function(void)
>>{
>> static std::string book_title;
>> Book any_book;
>> book_title = any_book.get_title();
>> return;
>>}
>>
>>I am throwing this code into a library,
>
>
> So now we can't know the full context, which imo
> is very likely relevant.
In other words, there is a third party involved:
source -> compiler -> librarian -> linker.
Unix libraries and builds have always been a
thorn for me.
>>and the library
>>reports:
>> U global destructors keyed to Global_Book
>> U global constructors keyed to Global_Book
>
>
> The 'library' reports? Or a compiler, or a linker?
> I don't understand.
Actually, this is a report generated from the library.
The 'U' indicates undefined symbol in the library.
In Cygwin / unix terms:
nm --demangle my_library.a
>>If I comment-out the global declaration, these messages
>>disappear {of course}.
>
>
> I don't know what those messages about 'keyed' mean.
> What does your documentation say?
Yep, that's what I'm having an issue with.
I was wondering if the C++ language specification had
any information about construction of global objects
and how it differs from construction of function-local
or static variables in functions.
>>So, why are there constructors & destructors for a global
>>instance, but not for an instance local to the function?
>
>
> Must be some propery of the library, or your library manager.
> A wild guess: Since I suspect you've only showed a 'skeleton'
> example, I'll ask: do any of your global objects depend upon
> other global objects for construction, in a particular order
> (the language does not specify this order for global objects
> between translation units.
My understanding is that constructors are called for all
objects. Just the order of construction and residence of
the object are the difference between global, function-local
and function-static variables.
>>If I change the declaration of "any_book" in the function
>>to static, there are still no "global constructors" listed
>>for it (I would think there would be).
>
>
> 'static' does not mean 'global'. Inside a function, it
> indicates a storage duration, nothing to do with scope.
Understood. A variable declared as static within a function
has the same life time as a global variable. Many compilers
place these two kinds of variables in the same area. My
understanding is that the global variable is constructed
sometime before "main" and the function-static variable is
construction upon the first entry to the function (the basis
of the Singleton pattern}.
>>I am using g++ (GCC) 3.3.1 (cygming special).
>
>
> I think you might have a platform-specific or tool-specific
> issue, but I can't say conclusively without more context.
> Have you tried any methodical, incremental build-and-testing?
>
> -Mike
The issue came about because I had a static constant array
of Books. I narrowed down to just one global instance
(removing the module static declaration). This gives the
error. However, sticking the variable inside a function
does not generate the error. This is the basis for
my issue. I have instances of this variable in many
functions. Its only the global declaration that is the
problem.
--
Thomas Matthews
C++ newsgroup welcome message:
http://www.slack.net/~shiva/welcome.txt
C++ Faq: http://www.parashift.com/c++-faq-lite
C Faq: http://www.eskimo.com/~scs/c-faq/top.html
alt.comp.lang.learn.c-c++ faq:
http://www.comeaucomputing.com/learn/faq/
Other sites:
http://www.josuttis.com -- C++ STL Library book
http://www.sgi.com/tech/stl -- Standard Template Library
- Next message: Aslan Kral: "Re: Good use for Unions"
- Previous message: Andre Kostur: "Re: Good use for Unions"
- In reply to: Mike Wahler: "Re: Variable declaration: Global vs. Function"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
