mingw-users Mailing List for MinGW - Minimalist GNU for Windows

On 2005-7-31 17:19 UTC, stefano luceri wrote:
> I don't know if it's the appropriate mailing-list ... if not I'll
> appreciate any advice.
> 
> I've the following problem writing a simple program. I would use the
> metods of std::valarray class adding some self writed method for elemnts
> of double type.

What extra functions do you want to define? It may be better
to write them as free functions instead of member functions.

> I've thought something like this:
> 
> /****************************/

#include <valarray>

> using namespace std;
> 
> class mvector : public valarray<double>

Class template std::valarray<T> has a nonvirtual destructor,
so you shouldn't derive from it.

> {
> mvector();
> mvector(int size);
> }

You need a ';' at the end of the class definition.

All members are private, which probably is not what you want.
The compiler should warn about that.

> mvector :: mvector() : valarray<double>() {}
> mvector :: mvector(int size) : valarray<double>(size) {}

valarray declares this constructor:
  explicit valarray(size_t);
Why use a different signature for your class?

> int main()
> {
>    mvector v(3);
>      v = (v+2.0);
> }
> 
> /*****************************/
> 
> Compiling this code I obtain the following error:
> 
> geom.cpp: In function `int main()':
> geom.cpp:38: error: no match for 'operator=' in 'v = std::operator+(const
>   std::valarray<_Tp>&, const _Tp&) [with _Tp = double]((&1.0e+0))'
> geom.cpp:22: error: candidates are: mvector& mvector::operator=(const
> mvector&)
> make: *** [geom.o] Error 1
> 
> Why can't I use the operator = like I do with valarray?

valarray defines six copy assignment operators.

Your class doesn't explicitly define any, so this one
  mvector& mvector::operator=(mvector const&)
is implicitly defined. That's the only 'candidate' the compiler
finds, and it doesn't match the argument given in main().

I don't know if it's the appropriate mailing-list ... if not I'll 
appreciate any advice.

I've the following problem writing a simple program. I would use the 
metods of std::valarray class adding some self writed method for elemnts 
of double type.

I've thought something like this:

/****************************/
using namespace std;

class mvector : public valarray<double>
{
mvector();
mvector(int size);
}

mvector :: mvector() : valarray<double>() {}
mvector :: mvector(int size) : valarray<double>(size) {}

int main()
{
   mvector v(3);
     v = (v+2.0);
}

/*****************************/

Compiling this code I obtain the following error:

geom.cpp: In function `int main()':
geom.cpp:38: error: no match for 'operator=' in 'v = std::operator+(const
  std::valarray<_Tp>&, const _Tp&) [with _Tp = double]((&1.0e+0))'
geom.cpp:22: error: candidates are: mvector& mvector::operator=(const 
mvector&)
make: *** [geom.o] Error 1

Why can't I use the operator = like I do with valarray?

I don't know if it's the appropriate mailing-list ... if not I'll 
appreciate any advice.

I've the following problem writing a simple program. I would use the 
metods of std::valarray class adding some self writed method for elemnts 
of double type.

I've thought something like this:

/****************************/
using namespace std;

class mvector : public valarray<double>
{
mvector();
mvector(int size);
}

mvector :: mvector() : valarray<double>() {}
mvector :: mvector(int size) : valarray<double>(size) {}

int main()
{
    mvector v(3);
   
    v = (v+2.0);
}

/*****************************/

Compiling this code I obtain the following error:

geom.cpp: In function `int main()':
geom.cpp:38: error: no match for 'operator=' in 'v = std::operator+(const
   std::valarray<_Tp>&, const _Tp&) [with _Tp = double]((&1.0e+0))'
geom.cpp:22: error: candidates are: mvector& mvector::operator=(const 
mvector&)
make: *** [geom.o] Error 1

Why can't I use the operator = like I do with valarray?

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3273409
By: arkay

I didn't want to open a ticket for a question, maybe it was something I did
wrong.  So which forum should I use?

Cheers,
Rainer


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3273198
By: tml1024

Sigh. This is so stupid it isn't even funny.

1) What's wrong with displaying a "Microsoft Visual C++ Runtime Library" title?
That *is* the official name of the MSVCRT DLL that mingw-built programs use,
after all. What difference is it to an end-user if an error message has a title
that talks about "Visual C++ Runtime Library", or "MingW[gcc] Runtime Library"?
End-users don't know what a runtime library is, even less what mingw is. Besides,
it's written "MinGW".

2) Surely you shouldn't use C++-stye comments in a header for C programs. A
header prescribed in the C standard, no less.

3) A call to assert() is supposed to be usable as a call to a void function.
Your assert() expands into a block. It's not the same thing.

4) You stomp over the namespace. You define the variables "assertstr" and "res"
inside a block in your assert(). So what happens the program also happens to
have a variable "res", and calls assert(!res)?

5) You need parens around the EXP. Consider what happens if the program calls
assert(0 && 1).

6) You sprintf() something of unknown length to a fixed buffer.

7) Where do the MessageBox, MB_*, and PostQuitMessage identifiers come from?
Why should a compilation unit that includes the C standard <assert.h> have to
include some Windows-specific header? (Making assert.h include windows.h itself
would be even worse, of course.)

8) A call to assert() with a nonzero argument is supposed to terminate execution
*right there*. Not some time later if and when the application happens to notice
a WM_QUIT message it posts to itself. And I don't think giving the end-user
the chance to ignore the assertion is a good idea.

9) Have you ever heard of non-GUI programs? They can call assert(), too, and
PostQuitMessage() will do absolutely nothing for them.

 

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3273195
By: pallav_nawani

Hi all,

I have developed a chess program, and it sometimes crashes. This happens very
infrequently, so I have not been able to debug it. Is there a way I could cause
a core dump in case of a crash? I use win98 + Gcc 3.xx.

Thanks,
Pallav Nawani
IronCode Software


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=290275

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3273171
By: aliimran

this version checks if debugging 'NDEBUG' is set or not and adds code
respectively.


#ifndef ASSERT_H
#define ASSERT_H

#ifdef NDEBUG /*do nothing */
	#define assert(x)	((void)0)
#else 

#define assert(EXP) {  char assertstr[1024];  int res=0;  if(!EXP) {      sprintf(assertstr,"Assertion failed!\r\n\r\nFile : %s\r\nLine
: %ld\r\n\r\nExpression : %s",__FILE__,__LINE__,#EXP);      res = MessageBox(NULL,assertstr,"MingW[gcc] Runtime Library",0 + MB_ICONHAND
+ MB_ABORTRETRYIGNORE +  MB_DEFBUTTON3);	if(res == IDABORT) PostQuitMessage(0);  }}
#endif

#endif

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3273165
By: aliimran

The actual assert.h dispatched with mingw header files displays Microsoft Visual
C++ default message when assertion fails alongwith message title as "Microsoft
Visual C++ Runtime Library". I have written a new assert.h which performs same
action but shows 'mingw assertion failed' message.

Advanced developers and package maintainers: please let me know if it is right
or fix it if it needs soem changes. Becuase it does not include clean up of
Message Queue for WM_QUIT.

Please save below code as assert.h and replace with existing one in your include
folder.

regards
Ali Imran

///////////////////////////////////////////////////////////////////////
#ifndef ASSERT_H
#define ASSERT_H

//MingW native assertion

#define assert(EXP) {  char assertstr[1024];  int res=0;  if(!EXP) {      sprintf(assertstr,"Assertion failed!\r\n\r\nFile : %s\r\nLine
: %ld\r\n\r\nExpression : %s",__FILE__,__LINE__,#EXP);      res = MessageBox(NULL,assertstr,"MingW[gcc] Runtime Library",0 + MB_ICONHAND
+ MB_ABORTRETRYIGNORE +  MB_DEFBUTTON3);	if(res == IDABORT) PostQuitMessage(0);  }}

#endif
///////////////////////////////////////////////////////////////////////

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

On Thu, 28 Jul 2005 20:55:57 -0400 (EDT)
rr...@cs... (Ross Ridge) wrote:

> > I"m trying to find some information about the use of a DBE-like (Double
> >Buffer Extension) in Windows. What i need is to use backbuffers created
> >on the video card memory.
>=20
> DirectDraw and Direct3D allow access to backbuffers created in video
> memory under Windows.
>=20
> >Or if i"ll have to use ActiveX, have anybody used ActiveX with mingw?
>=20
> You don't have to use ActiveX, though you do need know the basics of COM.
>=20
> 						Ross Ridge



Sorry, i'm confused.
DirectDraw and Direct3D are components of DirectX, aren't they? I don't eve=
n know what COM is... so i guess using DirectX is the better choice?
I found today those usefull links with tips for people who wants to use Dir=
ectX with MinGW:
http://www.spacejack.org/games/mingw/mingw-dx.html
http://www.spacejack.org/games/mingw/mingw-sdl.html
But now i'll start looking at these COM ... by the way, thanks for your hel=
p Ross.
=09
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Guilherme Castro Royer
AQX Instrumenta=E7=E3o
Site: www.aqx.com.br
Email: in...@aq...
Fone: (48) 2107-2724
Fone: (48) 333-2770
Fax:  (48) 333-3745
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3271118
By: earnie

Take a look around the MinGWiki and Google.

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=7134

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3271058
By: asegura

I compiled an application (7zip's filemanager) with MinGW 4.1 (gcc 3.4.2), using
-Os, and striped it (there's no debug info).
The size of the exe is 725 KB.   (with -O6 I think it was over 900 KB)
The officially distributed exe, compiled with MSVC 6, is 260 KB !!
 
Both have exceptions enabled, as the program uses them (very seldom, though).
Both use the same source files and resources, and link to the almost same DLLs.
 
Is this behavior known? That's more than twice the size!
 
Also exception support tends to make the code grow more than MSVC.
 
What can be done?
Will gcc 4.* make the code smaller?
 

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=7134

Greetings,

>> Maybe you are allocating too much memory at a time? Malloc is suitable
>> for relatively small pieces of memory only. Use calloc and see whether
>> the problem disappears.

anc> What makes you think so?  `calloc' is really often implemented on top =
of
anc> `malloc', and the difference is only the interface and that calloc will
anc> clear all content to zeros, the latter of which is not wanted generall=
y.

anc> Best regards,

anc> Yongwei

I apologize. It appears to be not true. I've thought about malloc so for a =
long
time, don't know why... And in the documentation there is nothing about it.=
 So
I'm mistaking. Thanks for clearing the question.

---------
With best regards,
 yGREK heretix <he...@ya...>          http://www.i.com.ua/~richy=20
 ICQ UIN: 292696772
 Jabber ID : yG...@ja...

yGREK Heretix wrote:

>>>  Any help or comments concerning the following problem will be
>>>  greatly appreciated.  Basically, I am developing a scientific
>>>  application which uses Fortran 90, C, and C++ code, and does some
>>>  heavy operations on matrices.  The details are unimportant (I
>>>  think), as the program runs to completion without issues using 3D
>>>  matrices of smallish sizes.  When the matrix dimensions are
>>>  increased to say, 128 x 128 x 128, it appears as if I am running
>>>  out of memory...malloc() is failing and responding with errno 12
>>>  which maps to not enough memory.  The problem is that according to
>>>  every way I have been monitoring the memory of my system, I have
>>>  plenty left!  If anyone can come up with any explanation, or even
>>>  just something to investigate, I would very much appreciate it...I
>>>  am running on Windows XP with MinGW and Msys, using G95 to compile
>>>  and link, and using g++/gcc to compile the c++/c code.  Thanks so
>>>  much for everyones time.
> 
> Maybe you are allocating too much memory at a time? Malloc is suitable
> for relatively small pieces of memory only. Use calloc and see whether
> the problem disappears.

What makes you think so?  `calloc' is really often implemented on top of
`malloc', and the difference is only the interface and that calloc will
clear all content to zeros, the latter of which is not wanted generally.

Best regards,

Yongwei

> I"m trying to find some information about the use of a DBE-like (Double
>Buffer Extension) in Windows. What i need is to use backbuffers created
>on the video card memory.

DirectDraw and Direct3D allow access to backbuffers created in video
memory under Windows.

>Or if i"ll have to use ActiveX, have anybody used ActiveX with mingw?

You don't have to use ActiveX, though you do need know the basics of COM.

						Ross Ridge

-- 
 l/  //	  Ross Ridge -- The Great HTMU
[oo][oo]  rr...@cs...
-()-/()/  http://www.csclub.uwaterloo.ca/u/rridge/ 
 db  //	  

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3270265
By: asegura

OK, that works but it is a workaround for a problem: --enable-stdcall-fixup
option is broken or incomplete. I've seen this problem for years, and I reported
it, but it has been the same in all new versions of ld.

The documentation says:

 `--enable-stdcall-fixup'
 `--disable-stdcall-fixup'
      If the link finds a symbol that it cannot resolve, it will attempt to
do "fuzzy linking" by looking for another defined symbol that differs only in
the format of the symbol name (cdecl vs stdcall) and will resolve that symbol
by linking to the match.  For example, the undefined symbol `_foo' might be
linked to the function `_foo@12', or the undefined symbol `_bar@16' might be
linked to the function `_bar'.  When the linker does this, it prints a warning,
since it normally should have failed to link, but sometimes import libraries
generated from third-party dlls may need this feature to be usable.  If you
specify `--enable-stdcall-fixup', this feature is fully enabled and warnings
are not printed.  If you specify `--disable-stdcall-fixup', this feature is
disabled and such mismatches are considered to be errors.

In tests i did long ago I think the linker could link references as "name" to
existing symbol "name@N", but not "name@N" to "name", which is more useful.
(not really sure now)

For instance the Win32 API consists mainly of stdcall functions whithout @N
decorations. Therefore, if the fuzzy linking would work as documented, there
would be no need for import libraries, and programs could be linked directly
to the system DLLs.

IMHO that would save a lot of space in MinGW's Win32 API package. Wouldn't it?


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=7134

Hi guys!

I'm trying to find some information about the use of a DBE-like (Double Buf=
fer Extension) in Windows. What i need is to use backbuffers created on the=
 video card memory.=20
Does anybody knows if its possible to use DBE under mingw?
Or if i'll have to use ActiveX, have anybody used ActiveX with mingw?

Thanks for your help

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Guilherme Castro Royer
AQX Instrumenta=E7=E3o
Site: www.aqx.com.br
Email: in...@aq...
Fone: (48) 2107-2724
Fone: (48) 333-2770
Fax:  (48) 333-3745
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3270005
By: rickg22

>As a result, the value of the GPL is depreciated. 
>That's why I'm making an issue of this. 
> 
>Danny

OK, Danny, I agree. We posted downloads (they're sourceforge downloads from
the mingw directory) of the mingw source code tarballs to our downloads page.
I think that leaves the GPL issue settled :)


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3268725
By: fidler

I do. I offer our thing in several varietes. I just want users to save downloading
and installing - if they want to use mingw, they can download one package, install
it in single step and have it all setup, ready to run.

Separate downloading is just more complicated. Also, we are using faster ld
replacement, so that it is not 100% mingw (we supply code for this faster linker
with each installation with BSD license - it is written from the scratch, does
not contain any original code).

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3268714
By: jigal2

Hi,

Downloaded and used it on a present mingw distribution I have, and a VERY simple
project.

On the outset - very impressive.

More to the details:

Problems:
1. Adding libraries is too long.
    Should have an option to add libraries in a bunch.
    The message "add using lreative path" is 
    redundant after the 1st time.

2. I couldn't change the background color.
   It stayed white. Personally, I prefer gray.... 

3. I tried using the buildin support for gdb, and I 
    couldn't see any present line mark or anything.
   Trying to step a few times and the program 
    completed....
    In short - this does not seem to work.

    Also - where can I display variables?

Requests:
1. Find in files should support file pattern and 
    finding in foldrs as well - not just opened files
    or project files.
    (for instance, searching for a definition in  
     mingw's include folder....).

Good stuff:
1. Nice editor.
2. Very nice class information and code folding.
    Looks as powerful as eclipse...


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3268680
By: dannysmith

> How funny, because dev-cpp does NOT distribute the sourcecode for the MINGW
binaries it has. Dev-cpp has been around for years, and I haven't seen any complaint
about it. Are we having some license discrepancy problem?


IMO, Dev-Cpp is violating the GPL.  I have posted a message to that effect to
the Dev-cpp users list, but have not received any response from the Dev-Cpp
developers.

If you really want an authoritative answer, correspond with the copyright holders
for gcc, binutils, gdb, the libgmon.a code in the mingw runtime, etc   and ask
them what you  need to do.

Not making any attempt to fulfill GPL obligation sets a very bad example for
other windows projects ("Dev-Cpp doesn't care about the GPL, CodeBlocks doesn't
care, etc, so why should I?". As a result, the value of the GPL is depreciated.
That's why I'm making an issue of this.

Danny

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3268566
By: ross_ridge

>Anyway, I hate doing stupid things. I think
>copying files from one sf.net ftp folder to another
>will not help anything

I'd say SourceForge counts as the same place.  But if you think copying the
source is stupid, why are you copying binaries?  Why not just direct people
to the MinGW site in order to get binaries in the first place?

                       Ross Ridge


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3268525
By: rickg22

I posted on the wrong place.

(See my second post in this topic)
 
One thing to clarify. In codeblocks, the executable installer can _ONLY_ be
downloaded from sourceforge (we'd be crazy to use our own bandwidth for that,
anyway). Does this change the distribution terms regarding MINGW binaries?

We're also considering to post links to the mingw sourcecode tarballs in our
website, next to the source code downloads. Will this be OK to comply with the
GPL? Or we'd have to release some duplicate tarballs instead?

Thanks.

______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

[description of testcase snipped]

> If I run test.x, compiled and linked as shown above, the mallocs fail fai=
rly
> quickly.  If I compile and link as shown below though, I can malloc more,
> although it still fails with plenty of memory left available:
>=20
> g95 -o test.x MatrixTest.f90 MatrixTestFromF90.o -lstdc++ -Wl,--heap,0x20=
000000
>=20
> If I compile and link with -Wl,--heap,0x10000000, it gets even farther (w=
hy??).=20
> So it appears as if when I give the heap more memory, I can get over this
> hurdle.  The strange part about this is, when I increase the heap with my
> program (my real one, not this test example) it fails during a dynamic me=
mory
> allocation even earlier on...any ideas?  Thanks again.

malloc allocs from the heap so having a large heap should allow you
to malloc large amounts of memory...

Contrary to what another poster wrote malloc is perfectly capable
of allocating memory in large chunks (apart from that the testprogram
mallocs memory in chunks of 1kB -- nothing we should have to worry
even if said poster were correct).

I have no real practical experience with g77 or g95 (last Fortran **
program I wrote is ages ago).

However (wild guess):
Could it be that the g75 frontend adds some options regarding
stack/heap/static memory which effectively results in using up
one of these (e.g. heap) "before time" ?
(I seem to recall that Fortran uses lots of static and stack memory).

While googling for 'heap+fortran' I came across some notes
hinting towardsindicating possible problems here.

Proposed were things like adding '-fno-automatic'.

Also there was a remark regarding automatically putting stack and
heap into the same memory area (i.e. trying to _NOT_ do that).

Since you use malloc you probably don't have large static areas...

Anyway:
I suppose this is more a code generation/linker problem specific
to your environment since I just tried you modified (as you
described) testcase under Linux using g77 3.3.5 as well as
gfortran-4.0.2 and both work nicely.

Best,
Michael
=2D-=20
 Vote against SPAM - see http://www.politik-digital.de/spam/
 Michael Gerdau       email: mg...@te...
 GPG-keys available on request or at public keyserver

Read and respond to this message at: 
https://sourceforge.net/forum/message.php?msg_id=3268263
By: fidler

I have to admit I am guilty as well.

upp.sf.net

also ships with mingw and also does not provide GCC sources.

Anyway, I hate doing stupid things. I think copying files from one sf.net ftp
folder to another will not help anything.

Ok, perhaps I will provide link in Downloads page. Technically there is little
difference in providing link to sources there or in Files page.


______________________________________________________________________
You are receiving this email because you elected to monitor this forum.
To stop monitoring this forum, login to SourceForge.net and visit: 
https://sourceforge.net/forum/unmonitor.php?forum_id=286529

Hi Guys,

I thought I fixed this but it seems to be cropping up again and maybe
I configured the compiler wrong but the binaries are a bit different
still.

After compiling the final executable the one on Linux has an extra
section called .ctors while the executable on Windows does not have
this section. Any ideas? There are probably some differences in the
COFF files too but I am not able to check them at this time.

Thanks in advance,

Devin