Building RTEMS/GANT under Cygwin

[Pat Wells]

Hello,

I have the latest Cygwin release (1.1.8) installed on a Windows NT machine.  
Is it possible (realistic) to build RTEMS/GNAT in this environment?  
Attempting to build with a target of powerpc, hosted on Windows NT.

I've read the rtems mail archives and am aware of the cygwin-rtems tarball 
and install executable that David Fiddes created.  Was hoping that the 
latest Cygwin release would address whatever problems existed previously.  
But perhaps this is not the case.  What were the problems anyway? I couldn't 
discover that from the archives.

Prior to launching the bit_ada script, I added the path of my GNAT
executables to the head of PATH variable:

export PATH=/cygdrive/c/gnat/bin:$PATH

And am consequently using the "build tools" that came with the GNAT
distribution rather than the Cygwin ones (make, ld, as).  This issue has
caused me a little consternation, and as a consequence I have experimented 
with adding a --host=i686-pc-mingw32 parameter to the "configure" command 
appearing in the bit_ada script.  I did this because the GNAT distribution 
tools indicated they were configured for mingw32.

When attempting to perform the build (via the bit_ada script), I've
encountered problems where apparently the symbolic links can't be resolved 
properly.  When the bit_ada script bombs out due to one of these symbolic 
link problems, I have replaced the symbolic link with a file copy.  And in 
that manner have incrementally progressed throught the bit_ada script (a 
modified bit_ada script such that the "build one pass" activity is skip on 
all but the first attempt).  Finaly got to a point where very many errors 
were produced dealing with unknown opcodes.

Not feeling very confident about this approach, I later tried to modify the 
make_one_tree function in the common.sh file such that the symbolic linking 
was replaced by either hard links, or in the case of directories, a 
recursive copy.  But there still remained the symbolic links created by the 
configure and makefile scripts.

Is this a hopeless endeavor, or not?  Any suggestions?

Also, I noticed that for the snapshot releases, under ada-tools/sources,
there are two "diff" files for gdb. One has a file name of
"gdb-4.17-rtems-gnat-3.13p-20000429.diff" and the other
"gdb-4.17-rtems-gnat-3.13p-20000918.diff".  Assuming I ever get this far,
are both of these diff's to be applied, or just one?

Thank you,

Pat Wells

_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com