Jamey Hicks wrote:
> Cross-compiling is often a headache.
>
> Hmm, we need some nice xscale machines on the internet for the development
> platform.
I can't wait to get my hands on the xscale...
The skiffcluster is a lifesaver, but cross compiling on the intel/athlon
boxes is so much faster, once you get past the configuration hurdles.
I've been toying with hacking autoconf as part of my squash the gtk
project. Basically, in order to support cross compilation better in a
linux->linux environment, all we have to do is add rsh support to
autoconf, and have it "do the right thing" when presented with the
proper cross development options - to run the test binaries remotely.
add 4 environment variables to autoconf:
cross_compile=Y # standard configure option
cross_compile_host=remote_machine_name #
cross_compile_dirpath= # path to the mounted from the host development
tree on the target
cross_compile_basename= Y/N # append the current basename to the
cross_compile_dirpath?
cross_compile_rsh = ssh #
It turned out rsh to did not reliably return the exit status of the last
command executed. But ssh does. So I built ssh last week. Ssh is huge,
but with a little more NFS trickery it should run mounted on
/usr/local/.
Now on to hacking autoconf and m4. Is there anything fatally wrong with
this idea?
Received on Wed Oct 11 17:23:41 2000
This archive was generated by hypermail 2.1.8 : Tue May 04 2004 - 09:43:44 EDT