Stefan Seefeld <[EMAIL PROTECTED]> writes:
> Andreas Beck wrote:
> >
> > > What I did, is to allow for a separate build directory, from which you
> > > $call Berlin/configure, to set up a *local* build tree (Makefiles,
> > > $generated headers, etc.),
> >
> > This does work with the GGI libs. Look at the buildall script. It does
> > exactly this.
>
> yeah, I know it builds in a separate directory. But doesn't it require
> one part to be installed before the next can be linked ?
No, but you have to use the --with-uninst-gii flag twice, to point out
both the source directory and the build directory.
> > This is impossible due to the dynamic nature of LibGGI (the configfiles
> > would need considerable tweaking as would the whole config reading stuff in
> > order to find everything in the build tree which has a somewhat other layout
> > than an installed version), but it is easily possible to set $prefix to
> > something like $HOME/ggitest/installed and install there.
>
> hmm, that would probably improve things quite considerably. I know what you
> mean, locating the resources (modules, font files, etc.) is tricky. What I
> did, is to provide a resource file that contains paths. This file is provided
> to the berlin server at run time, so you can just generate one (with configure)
> that is suitable for the temporary environment.
Paths to all sublibs are specified in libggi.conf (and the equivalent
for other libraries), so you can just change the ".root:" line at the
top to use a different set of sublibs.
The path to the config file is compiled into the main library, but it
is prefixed with the string "pAtHTAg" and has unused padding at the
end, so it can easily be changed with the tool libggi/patchlib.
Only problem left is that libtool likes to compile applications with
the -rpath flag, which can't always be overriden by LD_LIBRARY_PATH,
so the demos will still use the main library found in $prefix/lib.
//Marcus
--
-------------------------------+------------------------------------
Marcus Sundberg | http://www.stacken.kth.se/~mackan
Royal Institute of Technology | Phone: +46 707 452062
Stockholm, Sweden | E-Mail: [EMAIL PROTECTED]
