Re: GCC 4.8.1 building fails

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Thank you both help. I'll use mingwbilds for now, I didn't knew about
them, and thought that GCC 4.6 is the latest version available, which
is pretty old. I also thought, that from all the ways to use gcc on
windows, MinGW would work the best.
Now I'm wondering, what is the most supported and reasonable way to
compile gcc on windows? without any sdk's, consoles, editors, etc.
just plain good gcc and its components what would produce native
windows binaries what do not require some special dlls like cygwin
does.

On Sat, Jun 15, 2013 at 2:53 PM, Jonathan Wakely <jwakely.gcc@xxxxxxxxx> wrote:
> On 15 June 2013 09:44, Ignas wrote:
>> I uninstalled MinGW and installed it again with all the latest stuff.
>> I tried to compile gcc 2 more times, without any luck. I defined
>> constants with paths to MinGW's include and lib folder, I tried to use
>> configure closely resembling the one used to compile gcc what came
>> with MinGW:
>> configure --disable-sjlj-exceptions --enable-shared --enable-libgomp
>> --disable-win32-registry --enable-lib-stdcxx-debug
>> --enable-version-specific-runtime-libs --build=mingw32 --prefix=/mingw
>> It always fails when it gets to a part of making things inside a build
>> folder, previously it was [objdir]/i686-pc-mingw32, this time it was
>> [objdir]/mingw32. Second time, I've tried to compile with binutils
>> source inside [srcdir].
>>
>>
>> Based on all that, I conclude, that it is a bug with gcc in area
>> related to Windows, MinGW, or compiling itself.
>> As I've said in my first email, gcc configures and makes all
>> libraries(gmp, mpfr, mpc, etc.) successfully, I've checked
>> [objdir]/[library name], they have .a, .o, etc. sometimes .exe and
>> .dll files, indicating what my MinGW installation is working fine, I'm
>> pretty sure it would have never compiled all that successfully if it
>> was broken. I've also checked [objdir]/gcc , it was also compiled
>> successfully.
>> Problem always starts somewhere when it tries to make the
>> [objdir]/[build folder], it appears to me, that at this point, gcc is
>> using its own compiled binaries to compile further, and that's when it
>> crashes.
>
> When I tried to build MinGW the only way I could make it work was
> using an "identity mount", I don't know why that's necessary. The
> whole experience of trying to build MinGW was extremely annoying and
> painful.  Quoting from
> http://mingw-users.1079350.n2.nabble.com/How-to-build-GCC-on-MinGW-MSYS-td5868612.html:
>
>
> These days, I make a point of building GCC in "identity" mounts, where
> the Unixy path ("/mingw") directly corresponds to the underlying Windows
> path ("C:\mingw"). This shouldn't be strictly necessary, but I suspect
> there are some gotchas that can be avoided that way. To change your
> /mingw to an identity mount, move E:\Temp\mingw\msys\1.0\mingw to
> E:\mingw, and edit your E:\Temp\mingw\msys\1.0\etc\fstab accordingly.




[Index of Archives]     [Linux C Programming]     [Linux Kernel]     [eCos]     [Fedora Development]     [Fedora Announce]     [Autoconf]     [The DWARVES Debugging Tools]     [Yosemite Campsites]     [Yosemite News]     [Linux GCC]

  Powered by Linux