Re: Importing GNU Classpath 0.91 on gcc classpath svn branch

[Mark Wielaard <mark at klomp dot org>]

Hi,

On Thu, 2006-05-18 at 10:34 -0600, Tom Tromey wrote:
> >>>>> "Mark" == Mark Wielaard <mark@klomp.org> writes:
> Mark>   - Other files should have a "GCJ LOCAL" comment, or are mentioned
> Mark>     in the classpath/ChangeLog.gcj file.
> Mark>     (Don't forget to svn resolved files.)
> 
> I think we ought to have a policy of no divergences underneath the
> classpath/ directory.  I think we fail this right now, but we can fix
> that.

It is mostly in the build files:
Makefile.am:## GCJ LOCAL: we remove 'tools' for the time being.
Makefile.am:## GCJ LOCAL: we need an extra -I here.
doc/Makefile.am:## GCJ LOCAL: we don't want to install Classpath's info
files.
include/Makefile.am:## GCJ LOCAL: headers are installed specially.
lib/Makefile.am:## GCJ LOCAL: Comment this out so we don't make an empty
                ## '$(glibjdir)' in the install tree.
native/jni/qt-peer/Makefile.am:## GCJ LOCAL: don't install this library
native/jni/gtk-peer/Makefile.am:## GCJ LOCAL: don't install this library
native/jawt/Makefile.am:## GCJ LOCAL: don't install this library

And two small fdlibm header changes:
native/fdlibm/fdlibm.h:/* GCJ LOCAL: Include files.  */
native/fdlibm/namespace.h:/* GCJ LOCAL We don't use the renaming here,
so this file is empty for us.

> Mark>   Possibly update the gcj/javaprims.h file with scripts/classes.pl
> Mark>   (See HACKING, it can only be done after the first source->bytecode
> Mark>    pass has finished.)
> 
> I'm tempted to generate this during the build.

That would be convenient. Currently (for me) it is a little cargo-cult
building. I just know that I need to do the javaprims dance when all
byte code is generated, but the native build fails.

Cheers,

Mark

Attachment: signature.asc
Description: This is a digitally signed message part