Hello, all ... I searched in the old ActiveState message archive, but didn't find anything about this. When I compile my Boost experiments with gcc3.3, everything seems to be fine; when I compile them with gcc4, however, I get things like this (from the simple_ls example): pugsleybutt:~/devstuff/c++/boost jmzorko$ ./simple_ls usage: simple_ls [path] In directory: /Users/jmzorko/devstuff/c++/boost filesystem simple_ls(27587) malloc: *** Deallocation of a pointer not malloced: 0x2801b18; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug filesystem.cpp simple_ls(27587) malloc: *** Deallocation of a pointer not malloced: 0x2801d00; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug simple_ls simple_ls(27587) malloc: *** Deallocation of a pointer not malloced: 0x2801b00; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug simple_ls.cpp simple_ls(27587) malloc: *** Deallocation of a pointer not malloced: 0x2801ce0; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug 4 files 0 directories 0 errors pugsleybutt:~/devstuff/c++/boost jmzorko$ The Boost version is 1.31 and my platform is Mac OSX 10.4.6 (PowerPC), if that matters ... i'd like to use gcc4 if possible, can anyone tell me how to achieve this? Regards, John Falling You - exploring the beauty of voice and sound http://www.fallingyou.com
Hi John, Versions 1.33_1 and earlier do not support the new gcc4.x series, you'll have to wait until 1.34 is released, or try your luck with the cvs head for the time being. Cheers, -- Manfred 06.06.2006 jmzorko@mac.com wrote:
Hello, all ...
I searched in the old ActiveState message archive, but didn't find anything about this. When I compile my Boost experiments with gcc3.3, everything seems to be fine; when I compile them with gcc4, however, I get things like this (from the simple_ls example):
pugsleybutt:~/devstuff/c++/boost jmzorko$ ./simple_ls
usage: simple_ls [path]
In directory: /Users/jmzorko/devstuff/c++/boost
filesystem simple_ls(27587) malloc: *** Deallocation of a pointer not malloced: 0x2801b18; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug
[]
Versions 1.33_1 and earlier do not support the new gcc4.x series, you'll have to wait until 1.34 is released, or try your luck with the cvs head for the time being.
Is there a list of problem libraries somewhere? Fedora Core 4 comes with gcc 4.0.2 and boost 1.32, but I've not noticed any problems.
Manfred Doudar wrote:
Darren Cook wrote:
Versions 1.33_1 and earlier do not support the new gcc4.x series, you'll have to wait until 1.34 is released, or try your luck with the cvs head for the time being.
Is there a list of problem libraries somewhere?
Good question. ... Anyone?
To the best of my knowledge there should be virtually no changes in Boost to accomodate gcc-4, certainly there are no configuration changes needed. The original report looks like it might be either a bug in the Filesystem lib code, or else in gcc's libstdc++, but we'd need a backtrace of the problem deallocations to stand any chance of figuring out which. If you can come up with more information please repost with [filesystem] in the title so the relevent people see the message. Thanks, John.
On Tue, Jun 06, 2006 at 09:46:07AM +0100, John Maddock wrote:
The original report looks like it might be either a bug in the Filesystem lib code, or else in gcc's libstdc++, but we'd need a backtrace of the problem deallocations to stand any chance of figuring out which. If you can come up with more information please repost with [filesystem] in the title so the relevent people see the message.
My guess would be that he's linking with (library) code that is compiled with two different versions of g++. -- Carlo Wood <carlo@alinoe.com>
Carlo,
The original report looks like it might be either a bug in the Filesystem lib code, or else in gcc's libstdc++, but we'd need a backtrace of the problem deallocations to stand any chance of figuring out which. If you can come up with more information please repost with [filesystem] in the title so the relevent people see the message.
My guess would be that he's linking with (library) code that is compiled with two different versions of g++.
That's exactly what it was ... the Boost package I installed via Fink (a popular Mac OSX package manager) was 1.31, and it was a binary (likely compiled with gcc3.x). I realized this once I tried compiling my sample with gcc3.3 and it worked ... Regards, John Falling You - exploring the beauty of voice and sound http://www.fallingyou.com
John et al,
To the best of my knowledge there should be virtually no changes in Boost to accomodate gcc-4, certainly there are no configuration changes needed.
The original report looks like it might be either a bug in the Filesystem lib code, or else in gcc's libstdc++, but we'd need a backtrace of the problem deallocations to stand any chance of figuring out which. If you can come up with more information please repost with [filesystem] in the title so the relevent people see the message.
The latest Boost from CVS works fine under gcc4 for me, at least so far (running Mac OSX 10.4.6). Regards, John Falling You - exploring the beauty of voice and sound http://www.fallingyou.com
participants (5)
-
Carlo Wood -
Darren Cook -
jmzorko＠mac.com -
John Maddock -
Manfred Doudar