Thread: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x

Brought to you by: artyom-beilis

cppcms-users

[Cppcms-users] Compilation errors using gcc 4.5 and c++0x

From: Stanimir M. <sta...@zo...> - 2010-06-25 11:53:24

Hello Artyom,

Congratulations for the beta release of CppCms 1.x.x!

I am using the new features of gcc 4.4 and --std=c++0x

Here are some compile errors from it.

[  1%] Building CXX object
booster/CMakeFiles/booster.dir/lib/aio/src/socket.cpp.o
In file included from /usr/include/c++/4.4/bits/stl_algobase.h:66,
                 from /usr/include/c++/4.4/bits/char_traits.h:41,
                 from /usr/include/c++/4.4/string:42,
                 from /usr/include/c++/4.4/stdexcept:39,
                 from
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/function.h:11,
                 from
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/socket.h:12,
                 from
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/lib/aio/src/socket.cpp:28:
/usr/include/c++/4.4/bits/stl_pair.h: In constructor ‘std::pair<_T1,
_T2>::pair(_U1&&, _U2&&) [with _U1 = int, _U2 = int, _T1 = const
booster::aio::buffer_impl<char*>::entry*, _T2 = unsigned int]’:
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:34:
  instantiated from ‘std::pair<const
booster::aio::buffer_impl<Pointer>::entry*, unsigned int>
booster::aio::buffer_impl<Pointer>::get() const [with Pointer =
char*]’
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:77:
  instantiated from here
/usr/include/c++/4.4/bits/stl_pair.h:90: error: invalid conversion
from ‘int’ to ‘const booster::aio::buffer_impl<char*>::entry*’
/usr/include/c++/4.4/bits/stl_pair.h: In constructor ‘std::pair<_T1,
_T2>::pair(_U1&&, _U2&&) [with _U1 = int, _U2 = int, _T1 = const
booster::aio::buffer_impl<const char*>::entry*, _T2 = unsigned int]’:
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:34:
  instantiated from ‘std::pair<const
booster::aio::buffer_impl<Pointer>::entry*, unsigned int>
booster::aio::buffer_impl<Pointer>::get() const [with Pointer = const
char*]’
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/lib/aio/src/socket.cpp:377:
  instantiated from here
/usr/include/c++/4.4/bits/stl_pair.h:90: error: invalid conversion
from ‘int’ to ‘const booster::aio::buffer_impl<const char*>::entry*’
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/system_error.h:
At global scope:
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/system_error.h:56:
warning: ‘booster::system::posix_category’ defined but not used
make[2]: *** [booster/CMakeFiles/booster.dir/lib/aio/src/socket.cpp.o] Error 1
make[1]: *** [booster/CMakeFiles/booster.dir/all] Error 2
make: *** [all] Error 2

Greetings,
Stanimir

Re: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x

From: Artyom <art...@ya...> - 2010-06-25 12:51:21

Hello,

Small note - not 4.5 -- 4.4

To be honest, it looks like a compiler bug... Or I do not know C++.
The funniest thing that I did tested the code with gcc-4.5 and c++0x flag
and it worked well. (See http://cppcms.sourceforge.net/wikipp/en/page/cppcms_1x_platforms)

gcc-4.4 (that I fortunately have installed) fails on

   return buffer_data_type(0,0);

Where buffer_data_type is std:pair of <entry const *,size_t> and entry itself
defined as struct entry { Pointer ptr; size_t size; } where Pointer is template parameters.

Now gcc-4.4 complains that it can't convert 0 (int) to entry const *, where it is clearly (at least for me)
legal as I can assign 0 to pointer.

Now I fixed this in the code by creating empty buffer_data_type and manually setting first and second to 0.

But what I do interested is whether I'm wrong or compiler is wrong.

I've commited the fix in changeset 1296
You may take it from the svn or just apply these changes in your copy as shown there:

http://cppcms.svn.sourceforge.net/viewvc/cppcms/framework/branches/refactoring/booster/booster/aio/buffer.h?r1=1173&r2=1296

Thanks for the report,
  Artyom




----- Original Message ----
> From: Stanimir Mladenov <sta...@zo...>
> To: cpp...@li...
> Sent: Fri, June 25, 2010 2:28:48 PM
> Subject: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x
> 
> Hello Artyom,

Congratulations for the beta release of CppCms 
> 1.x.x!

I am using the new features of gcc 4.4 and --std=c++0x

Here 
> are some compile errors from it.

[  1%] Building CXX 
> object
booster/CMakeFiles/booster.dir/lib/aio/src/socket.cpp.o
In file 
> included from /usr/include/c++/4.4/bits/stl_algobase.h:66,
    
>              from 
> /usr/include/c++/4.4/bits/char_traits.h:41,
        
>          from /usr/include/c++/4.4/string:42,
  
>                from 
> /usr/include/c++/4.4/stdexcept:39,
            
>     
> from
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/function.h:11,
  
>               
> from
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/socket.h:12,
  
>               
> from
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/lib/aio/src/socket.cpp:28:
/usr/include/c++/4.4/bits/stl_pair.h: 
> In constructor ‘std::pair<_T1,
_T2>::pair(_U1&&, _U2&&) 
> [with _U1 = int, _U2 = int, _T1 = 
> const
booster::aio::buffer_impl<char*>::entry*, _T2 = unsigned 
> int]’:
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:34:
  
> instantiated from 
> ‘std::pair<const
booster::aio::buffer_impl<Pointer>::entry*, 
> unsigned int>
booster::aio::buffer_impl<Pointer>::get() const [with 
> Pointer 
> =
char*]’
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:77:
  
> instantiated from here
/usr/include/c++/4.4/bits/stl_pair.h:90: error: 
> invalid conversion
from ‘int’ to ‘const 
> booster::aio::buffer_impl<char*>::entry*’
/usr/include/c++/4.4/bits/stl_pair.h: 
> In constructor ‘std::pair<_T1,
_T2>::pair(_U1&&, _U2&&) 
> [with _U1 = int, _U2 = int, _T1 = const
booster::aio::buffer_impl<const 
> char*>::entry*, _T2 = unsigned 
> int]’:
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:34:
  
> instantiated from 
> ‘std::pair<const
booster::aio::buffer_impl<Pointer>::entry*, 
> unsigned int>
booster::aio::buffer_impl<Pointer>::get() const [with 
> Pointer = 
> const
char*]’
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/lib/aio/src/socket.cpp:377:
  
> instantiated from here
/usr/include/c++/4.4/bits/stl_pair.h:90: error: 
> invalid conversion
from ‘int’ to ‘const booster::aio::buffer_impl<const 
> char*>::entry*’
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/system_error.h:
At 
> global 
> scope:
/home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/system_error.h:56:
warning: 
> ‘booster::system::posix_category’ defined but not used
make[2]: *** 
> [booster/CMakeFiles/booster.dir/lib/aio/src/socket.cpp.o] Error 1
make[1]: 
> *** [booster/CMakeFiles/booster.dir/all] Error 2
make: *** [all] Error 
> 2

Greetings,
Stanimir

------------------------------------------------------------------------------
ThinkGeek 
> and WIRED's GeekDad team up for the Ultimate 
GeekDad Father's Day Giveaway. 
> ONE MASSIVE PRIZE to the 
lucky parental unit.  See the prize list and 
> enter to win: 
> 
http://p.sf.net/sfu/thinkgeek-promo
_______________________________________________
Cppcms-users 
> mailing list

> href="mailto:Cpp...@li...">Cpp...@li...

> href="https://lists.sourceforge.net/lists/listinfo/cppcms-users" target=_blank 
> >https://lists.sourceforge.net/lists/listinfo/cppcms-users

Re: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x

From: Stanimir M. <sta...@zo...> - 2010-06-26 09:46:20

Thanks, I fetched it from svn!

Yes, gcc 4.4 and not 4.5 has the auto and some other nice c++0x
features. I will switch in a day to 4.5 for more of them.

Actually, CppCms compiles fine without the --std=c++0x flag. I saw you
are using the auto keyword in the templates loops if available and
since I am using it too, decided to give it to CppCms also.
About the error, I think that because the 0 (int) argument is an
rvalue and converting it to non const pointer (because the return
result is non const).

On Fri, Jun 25, 2010 at 3:51 PM, Artyom <art...@ya...> wrote:
>
> Hello,
>
> Small note - not 4.5 -- 4.4
>
> To be honest, it looks like a compiler bug... Or I do not know C++.
> The funniest thing that I did tested the code with gcc-4.5 and c++0x flag
> and it worked well. (See http://cppcms.sourceforge.net/wikipp/en/page/cppcms_1x_platforms)
>
> gcc-4.4 (that I fortunately have installed) fails on
>
>   return buffer_data_type(0,0);
>
> Where buffer_data_type is std:pair of <entry const *,size_t> and entry itself
> defined as struct entry { Pointer ptr; size_t size; } where Pointer is template parameters.
>
> Now gcc-4.4 complains that it can't convert 0 (int) to entry const *, where it is clearly (at least for me)
> legal as I can assign 0 to pointer.
>
> Now I fixed this in the code by creating empty buffer_data_type and manually setting first and second to 0.
>
> But what I do interested is whether I'm wrong or compiler is wrong.
>
> I've commited the fix in changeset 1296
> You may take it from the svn or just apply these changes in your copy as shown there:
>
> http://cppcms.svn.sourceforge.net/viewvc/cppcms/framework/branches/refactoring/booster/booster/aio/buffer.h?r1=1173&r2=1296
>
> Thanks for the report,
>  Artyom
>
>
>
>
> ----- Original Message ----
> > From: Stanimir Mladenov <sta...@zo...>
> > To: cpp...@li...
> > Sent: Fri, June 25, 2010 2:28:48 PM
> > Subject: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x
> >
> > Hello Artyom,
>
> Congratulations for the beta release of CppCms
> > 1.x.x!
>
> I am using the new features of gcc 4.4 and --std=c++0x
>
> Here
> > are some compile errors from it.
>
> [  1%] Building CXX
> > object
> booster/CMakeFiles/booster.dir/lib/aio/src/socket.cpp.o
> In file
> > included from /usr/include/c++/4.4/bits/stl_algobase.h:66,
>
> >              from
> > /usr/include/c++/4.4/bits/char_traits.h:41,
>
> >          from /usr/include/c++/4.4/string:42,
>
> >                from
> > /usr/include/c++/4.4/stdexcept:39,
>
> >
> > from
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/function.h:11,
>
> >
> > from
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/socket.h:12,
>
> >
> > from
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/lib/aio/src/socket.cpp:28:
> /usr/include/c++/4.4/bits/stl_pair.h:
> > In constructor ‘std::pair<_T1,
> _T2>::pair(_U1&&, _U2&&)
> > [with _U1 = int, _U2 = int, _T1 =
> > const
> booster::aio::buffer_impl<char*>::entry*, _T2 = unsigned
> > int]’:
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:34:
>
> > instantiated from
> > ‘std::pair<const
> booster::aio::buffer_impl<Pointer>::entry*,
> > unsigned int>
> booster::aio::buffer_impl<Pointer>::get() const [with
> > Pointer
> > =
> char*]’
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:77:
>
> > instantiated from here
> /usr/include/c++/4.4/bits/stl_pair.h:90: error:
> > invalid conversion
> from ‘int’ to ‘const
> > booster::aio::buffer_impl<char*>::entry*’
> /usr/include/c++/4.4/bits/stl_pair.h:
> > In constructor ‘std::pair<_T1,
> _T2>::pair(_U1&&, _U2&&)
> > [with _U1 = int, _U2 = int, _T1 = const
> booster::aio::buffer_impl<const
> > char*>::entry*, _T2 = unsigned
> > int]’:
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/aio/buffer.h:34:
>
> > instantiated from
> > ‘std::pair<const
> booster::aio::buffer_impl<Pointer>::entry*,
> > unsigned int>
> booster::aio::buffer_impl<Pointer>::get() const [with
> > Pointer =
> > const
> char*]’
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/lib/aio/src/socket.cpp:377:
>
> > instantiated from here
> /usr/include/c++/4.4/bits/stl_pair.h:90: error:
> > invalid conversion
> from ‘int’ to ‘const booster::aio::buffer_impl<const
> > char*>::entry*’
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/system_error.h:
> At
> > global
> > scope:
> /home/stanimir/work/web_common/third_parties/cppcms/framework/branches/refactoring/booster/./booster/system_error.h:56:
> warning:
> > ‘booster::system::posix_category’ defined but not used
> make[2]: ***
> > [booster/CMakeFiles/booster.dir/lib/aio/src/socket.cpp.o] Error 1
> make[1]:
> > *** [booster/CMakeFiles/booster.dir/all] Error 2
> make: *** [all] Error
> > 2
>
> Greetings,
> Stanimir
>
> ------------------------------------------------------------------------------
> ThinkGeek
> > and WIRED's GeekDad team up for the Ultimate
> GeekDad Father's Day Giveaway.
> > ONE MASSIVE PRIZE to the
> lucky parental unit.  See the prize list and
> > enter to win:
> >
> http://p.sf.net/sfu/thinkgeek-promo
> _______________________________________________
> Cppcms-users
> > mailing list
>
> > href="mailto:Cpp...@li...">Cpp...@li...
>
> > href="https://lists.sourceforge.net/lists/listinfo/cppcms-users" target=_blank
> > >https://lists.sourceforge.net/lists/listinfo/cppcms-users
>
>
>
>
> ------------------------------------------------------------------------------
> ThinkGeek and WIRED's GeekDad team up for the Ultimate
> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the
> lucky parental unit.  See the prize list and enter to win:
> http://p.sf.net/sfu/thinkgeek-promo
> _______________________________________________
> Cppcms-users mailing list
> Cpp...@li...
> https://lists.sourceforge.net/lists/listinfo/cppcms-users

Re: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x

From: Artyom <art...@ya...> - 2010-06-26 19:44:21

> 
> Actually, CppCms compiles fine without the --std=c++0x
> flag. I saw you
> are using the auto keyword in the templates loops if
> available and
> since I am using it too, decided to give it to CppCms
> also.

Note, I don't really need C++0x mode to get nice way to iterate
over collections.

Under gcc, intel and even SunStudo (I used to support) 
compilers I have typeof or __typeof__ which is very close to decltype
so auto is not so necessary, but still I'm really looking forward
for it in all modern compilers as for C++0x at all.

If today C++0x was supported by most compilers I would not need
half of Booster today.

Regards
  Artyom

Re: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x

From: Stanimir M. <sta...@zo...> - 2010-06-28 09:07:08

Yes, but using C++0x gives much benefits already and it is really good
that you support it to some extend already.
Something that already was discussed, is the deprecation of auto_ptr
in favor of unique_ptr. At the moment c++0x is enabled,  a warning
about this is emitted. Is it possible to substitute auto_ptr to
unique_ptr the same way you do with auto and __typeof__?

It is not a big deal but it looks much better this way.

Stanimir

On Sat, Jun 26, 2010 at 10:44 PM, Artyom <art...@ya...> wrote:
>>
>> Actually, CppCms compiles fine without the --std=c++0x
>> flag. I saw you
>> are using the auto keyword in the templates loops if
>> available and
>> since I am using it too, decided to give it to CppCms
>> also.
>
> Note, I don't really need C++0x mode to get nice way to iterate
> over collections.
>
> Under gcc, intel and even SunStudo (I used to support)
> compilers I have typeof or __typeof__ which is very close to decltype
> so auto is not so necessary, but still I'm really looking forward
> for it in all modern compilers as for C++0x at all.
>
> If today C++0x was supported by most compilers I would not need
> half of Booster today.
>
> Regards
>  Artyom
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by Sprint
> What will you do first with EVO, the first 4G phone?
> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> _______________________________________________
> Cppcms-users mailing list
> Cpp...@li...
> https://lists.sourceforge.net/lists/listinfo/cppcms-users
>

Re: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x

From: Artyom <art...@ya...> - 2010-06-28 10:32:16

No it is not quite possible,

1. unique_ptr and auto_ptr are different pointers with different semantics.

   For example:

   auto_ptr<foo> factory()
   {
       auto_ptr<foo> ptr(new foo());
       return ptr;
   }

   Would not work with unique_ptr as it requires std::move.

2. They are different binary-incompatible classes. For example my code runs
   on gcc-4.5 without C++0x, if in two years C++0x becomes default with
   gcc then automatic replacement of auto_ptr with unique_ptr will break
   backward binary compatibility of the library, so I can't relay
   on this feature.

I think it is stupid to deprecate auto_ptr as it is the only smart pointer existing
in C++03 so I hope gcc will remove deprecation warning for most used C++ smart pointer.

I don't see any problem with it, the only problem is that some programmers
had used it inside STL Collections and got problems with it.


Artyom



----- Original Message ----
> From: Stanimir Mladenov <sta...@zo...>
> To: cpp...@li...
> Sent: Mon, June 28, 2010 12:06:38 PM
> Subject: Re: [Cppcms-users] Compilation errors using gcc 4.5 and c++0x
> 
> Yes, but using C++0x gives much benefits already and it is really good
that 
> you support it to some extend already.
Something that already was discussed, 
> is the deprecation of auto_ptr
in favor of unique_ptr. At the moment c++0x is 
> enabled,  a warning
about this is emitted. Is it possible to substitute 
> auto_ptr to
unique_ptr the same way you do with auto and 
> __typeof__?

It is not a big deal but it looks much better this 
> way.

Stanimir

On Sat, Jun 26, 2010 at 10:44 PM, Artyom <
> ymailto="mailto:art...@ya..." 
> href="mailto:art...@ya...">art...@ya...> 
> wrote:
>>
>> Actually, CppCms compiles fine without the 
> --std=c++0x
>> flag. I saw you
>> are using the auto keyword 
> in the templates loops if
>> available and
>> since I am using 
> it too, decided to give it to CppCms
>> also.
>
> Note, I 
> don't really need C++0x mode to get nice way to iterate
> over 
> collections.
>
> Under gcc, intel and even SunStudo (I used to 
> support)
> compilers I have typeof or __typeof__ which is very close to 
> decltype
> so auto is not so necessary, but still I'm really looking 
> forward
> for it in all modern compilers as for C++0x at 
> all.
>
> If today C++0x was supported by most compilers I would not 
> need
> half of Booster today.
>
> Regards
> 
>  Artyom
>
>
>
>
>
>
> 
> ------------------------------------------------------------------------------
> 
> This SF.net email is sponsored by 
> Sprint
> What will you do first with EVO, the first 4G phone?
> 
> Visit sprint.com/first -- 
> http://p.sf.net/sfu/sprint-com-first
> 
> _______________________________________________
> Cppcms-users mailing 
> list
> 
> href="mailto:Cpp...@li...">Cpp...@li...
> 
> 
> target=_blank 
> >https://lists.sourceforge.net/lists/listinfo/cppcms-users
>

------------------------------------------------------------------------------
This 
> SF.net email is sponsored by Sprint
What will you do first with EVO, the 
> first 4G phone?
Visit sprint.com/first -- 
> href="http://p.sf.net/sfu/sprint-com-first" target=_blank 
> >http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Cppcms-users 
> mailing list

> href="mailto:Cpp...@li...">Cpp...@li...

> href="https://lists.sourceforge.net/lists/listinfo/cppcms-users" target=_blank 
> >https://lists.sourceforge.net/lists/listinfo/cppcms-users