TDM-GCC 4.5 series (Latest: 4.5.2 - 2011-03-27)

[TDragon]

Can you confirm that 32-bits libs build with the 32/64-bits edition are compatibles with the 32-bits only edition ?

Yes. But you're responsible for knowing the ABI of any languages and libraries you try to use in this manner, so please don't post any questions or bug reports when you run into problems when doing this.

I've tried with the 2.8.12, and encountered problems :
...
If UNICODE=1, all works fine.

Is there something I missed, or is this confirmed ?

I haven't tried with 2.8.12.

[Xaviou]

Hi

Can you confirm that 32-bits libs build with the 32/64-bits edition are compatibles with the 32-bits only edition ?

Yes. But you're responsible for knowing the ABI of any languages and libraries you try to use in this manner, so please don't post any questions or bug reports when you run into problems when doing this.

Of course. Thanks for the response.

I've tried with the 2.8.12, and encountered problems :
...
If UNICODE=1, all works fine.

Is there something I missed, or is this confirmed ?

I haven't tried with 2.8.12.

Resolved the problem with UNICODE=0 by adding a #include <direct.h> line in src/common/filefn.cpp  and src/common/debugrpt.cpp (thank you reckless for helping me by PM).
So I now can confirm that wxWidgets-2.8.12 can succesfully be build with TDM-GCC 4.5.2 with all classics configurations (I've tried all the 16 differents combinations) :

• Release / Debug
• Ansi / Unicode
• Static / Dynamic
• Multilib / Monolithic

Thanks again to all

Regards

Xav'

[stahta01]

I was compiling code from http://cboard.cprogramming.com/cplusplus-programming/139494-getconsolescreenbufferinfoex-function.html#post1039300

And, I found,what I think are, a few typos in w64 headers.

Two in file wincrypt.h.
The structure define of CertSelectCertificateChains
Had "cCriteria DWORD," instead of "DWORD cCriteria,"

The structure define of CryptExportPublicKeyInfoFromBCryptKeyHandle
Had "void pvAuxInfo," instead of what I think was meant of "void* pvAuxInfo,"

One in file objbase.h
The prototype of CoGetApartmentType
Had something else instead of "HRESULT WINAPI CoGetApartmentType("

NOTE: I am just a C Programmer; So, you need to verify these are typos.

Tim S.

[xunxun]

I was compiling code from http://cboard.cprogramming.com/cplusplus-programming/139494-getconsolescreenbufferinfoex-function.html#post1039300

And, I found,what I think are, a few typos in w64 headers.

Two in file wincrypt.h.
The structure define of CertSelectCertificateChains
Had "cCriteria DWORD," instead of "DWORD cCriteria,"

The structure define of CryptExportPublicKeyInfoFromBCryptKeyHandle
Had "void pvAuxInfo," instead of what I think was meant of "void* pvAuxInfo,"

One in file objbase.h
The prototype of CoGetApartmentType
Had something else instead of "HRESULT WINAPI CoGetApartmentType("

NOTE: I am just a C Programmer; So, you need to verify these are typos.

Tim S.

You can use latest MinGW64 svn.
These have been fixed.

[TDragon]

Perhaps I'll create a new mingw64-runtime package in the near future.

[Loaden]

Perhaps I'll create a new mingw64-runtime package in the near future.

Hope to upgrade the latest of GCC!

[reckless]

im sure he will

in the meantime if you want to try the new gcc i made some builds public on my google code site.

http://mingw-wine.googlecode.com/files/mingw32-gcc-4.6.1-release.7z 32 bit mingw gcc
http://mingw-wine.googlecode.com/files/mingw64-gcc-4.6.1-release.7z 64 bit mingw gcc

both use the mingw 64 runtime so if you plan on upgrading from these you need that as well as binutils built for this arch.
ill make a complete pack soon. still hunting down some strange bug with xunxun in regards to the lto plugin.

[Justin Brimm]

I can barely wait until he releases a GCC 4.6.x build; the newest release of GCC has an entirely rewritten Objective-C runtime library that brings it up to Apple's specs, including Obj-C 2.0 support, garbage collection and synchronization. After having used TDM-GCC for so long though, other builds seem lackluster by comparison. I have to admit that I've been spoiled by the full 64 and 32 bit compilation support, as well as the inclusion of (so far) fully functional OpenMP 3.0 and pthreads.