Re: [jamboost] Re: [Boost-build] [V2] [Qt4] Progress

From: Jürgen Hunold (hunold_at_[hidden])
Date: 2005-12-08 02:38:28

Hi Alo !

On Thursday 08 December 2005 05:39, Alo Sarv wrote:
> > > Well I have to say I'm a totally incompetent when it comes to QT. And
> > > I can't even begin to test your patches. So could someone else please
> > > do some tests so I can be sure it's not going to cause havoc :-)
>
> Since my main interest in the topic is building the QT4 free edition
> with MSVC, I did some testing of the jamfiles on win32, both mingw and
> msvc8.0 configurations. Here's my findings and recommended fixes:

Thanks for testing !

> 1. The alias platform_sources-mechanism in corelib/Jamfile (and
> possibly elsewhere as well) doesn't seem to work. On windows, it's
> complaining that io/qprocess_unix.cpp doesn't exist. Commenting out
> the entire linux-version of platform_sources solves the problem.

Yes. I've already fixed this is my current version.

> 2. io/qabstractfileengine.cpp doesn't exist in the qt win32 source
> tarball, so it should probably be moved to linux-specific source
> listing.

Interesting. But it listed in the qmake Include-File corelib/io/io.pri as a 
platform independent source.

> 3. In corelib/Jamfile, you'll want to use
> <link>shared:<define>QT_MAKEDLL and <link>shared:<define>QT_DLL (fixes
> mingw-builds on win32). Probably similar change in other places.

Those are probably best put in src/Jamfile as they are likely needed for every 
library. I should have known this since they are needed in 3x as well...

> 4. For msvc, we need to copy mkspecs/win32-g++ to mkspecs/win32-msvc.
> So far, qtcore built with this correctly, but maybe we need to make
> some tweaks there later on.

I think this should be handled by specifying the right include-paths depending 
on on the toolset. I'll take a look at how to get the current toolset in the 
jamfile

> After this, qtcore technically compiles with both mingw (after
> commenting out the gcc4-specific flags) and msvc, however something
> seems to be broken related to mocing - from what I can see, mocing
> isn't done at all for some reason. There's no error/warning that I can
> find, it just complains about failure to include moc_qbuffer.cpp or
> qtimer.moc...

Right, you've tried a rather old version. Those are files which are moc-only 
headers and those are not supported by the current Qt4 toolchain.
I'll have a patch for this issue and an improved set of Jamfiles at home.

If you send me the changes you did (or the whole Jamfiles, I'll use ediff 
then ;-)) ), then I can prepare an improved set and post both the Jamfiles 
and the needed patches on list for review again.

And could you please sign up to the new mailing list 
"boost-build_at_[hidden]"
as the old list will be shut down on January, 1st.

Yours,

Jürgen

-- 
* Dipl.-Math. Jürgen Hunold ! Institut für Verkehrswesen, Eisenbahnbau
* voice: ++49 511 762-2529 ! und -betrieb, Universität Hannover 
* fax : ++49 511 762-3001 ! Appelstrasse 9a, D-30167 Hannover
* hunold_at_[hidden] ! www.ive.uni-hannover.de