2011/5/9 Ruben Van Boxem <[email protected]>:
> 2011/5/9 K. Frank <[email protected]>
>>
>> Hello Ruben!
>>
>> On Mon, May 9, 2011 at 3:34 AM, Ruben Van Boxem
>> <[email protected]> wrote:
>> > Op 9 mei 2011 00:16 schreef "K. Frank" <[email protected]> het
>> > volgende:
>> > ...
>> >> > On the other hand, why rebuild what you already have? Only reason
>> >> > would
>> >> > be
>> >> > optimization, but I wouldn't bother. You should be able to use the
>> >> > new
>> >> > GCC
>> >> > version to link the older libraries. Just use the new compiler for
>> >> > your
>> >> > current project, using prebuilt libs.
>> >>
>> >> My current "everyday" installation of Qt (and some other small
>> >> libraries)
>> >> was built with a tdm mingw32 gcc build:
>> >>
>> >>   g++ --version:  g++ (TDM-2 mingw32) 4.4.1
>> >>
>> >> I just sort of assumed that I wouldn't be able to mix my current
>> >> (tdm 4.4.1) Qt libraries with code compiled with the new compiler.
>> >> Do I understand you correctly that they should be compatible?
>> >> (By the way, my tdm gcc uses sjlj exceptions.)
>> >
>> > Ah yes, I believe tdm uses the MinGW.org runtime, so compatibility
>> > between
>> > my mingw-w64 based toolchains and yours would not be guaranteed.
>>
>> Thanks for the heads-up.  If and when I decide to upgrade, I will plan on
>> rebuilding Qt.
>>
>> > ...
>> > I would like to upload my Qt
>> > Builds, but they are very hard, if not impossible to get relocatable.
>>
>> I assume that by relocatable, you mean being able to move the Qt
>> installation from one location to another in the file-system directory
>> hierarchy.
>>
>> Yes, this issue bit me once, and I ended up rebuilding Qt as the most
>> convenient solution.
>>
>> I believe that there was a discussion on the Qt-interest list about this
>> (speaking from memory), and that the Qt developers' stance was that
>> this non-relocatability is a feature, and it didn't seems that there was
>> any likelihood it would be changed.  I honestly never understood the
>> reasoning behind this.
>>
>> > Perhaps someone knows of a way better than the google code project which
>> > tried to get it right some time ago?
>>
>> I seem to recall that in the list discussion, someone mentioned a tool
>> that relocates a Qt installation by editing the hard-coded path names
>> in the executables.
>>
>> > How does Nokia make its SDK installer?
>>
>> I also seem to recall someone saying that the Nokia installer worked
>> in essentially the same way as the above-mentioned tool: that the
>> executables had their hard-coded path names (or perhaps placeholders)
>> edited upon installation to descend from the desired root directory for
>> the installation.
>>
>> Again, this is from memory -- sorry if this is bogus information.
>>
>> I expect that a number of people would find it a valuable service, were
>> you to solve the relocation issue and make available your Qt builds.
>> (I know I would.)  But your gcc builds are already a huge contribution,
>> and probably the more important piece of the puzzle.
>
> Yeah, it's still the mess I'd though it was :). I read those discussions a
> long time ago, and frankly, their reasoning makes no sense at all. Windows
> != Unix, 'nuff said.
>
> Just FYI, I just rebuilt Qt 4.7 (git branch) with my GCC 4.6.1 (prerelease)
> toolchain and everything went fine. I also built Qt Creator and works just
> as nice as the old one I had compiled with Sezero's GCC 4.5.2 toolchain.
>
> Just remember to configure with (when you do, of course)
>
> -qt-style-windowsxp -qt-style-windowsvista -phonon
>
> The first two tell Qt you do have the headers for Windows XP+ window
> decorations and widgets (or else you'll get the ugly win95 style for all
> your Qt apps), the latter to tell Qt you've got the headers required for
> phonon+DShow backend.
>
>> Thanks again for your contributions and all of your help.
>
> No problem at all, I was just frustrated there was no non-autobuild of GCC
> 4.6 on the mingw-w64 site, and put myself to learning some Bash :P
>
> Ruben
>
>>
>> K. Frank
>>

Hmm, is it possible to get a stack-trace for ld's plugin crash? This
might be pretty helpful to know at what point it crashes.   As I don't
see this issue I assume it is mingw-.host only related, so a
stack-trace would be kind.

Regards,
Kai

------------------------------------------------------------------------------
Achieve unprecedented app performance and reliability
What every C/C++ and Fortran developer should know.
Learn how Intel has extended the reach of its next-generation tools
to help boost performance applications - inlcuding clusters.
http://p.sf.net/sfu/intel-dev2devmay
_______________________________________________
Mingw-w64-public mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public

Reply via email to