In many places, the configure scripts do tests like this:
case "$host" in ...
Should those be changed to:
case "$target" in ...
?
I find, in trying to cross-compile for arm-linux from i686-linux (for the Rio Receiver), that the configure doesn't "do the right thing", which is what leads to the fix-make script in Jeff Mock's patch here:
http://www.mock.com/receiver/demo/
I'd rather get the configure scripts "corrected" (if I am, in fact, right about the host vs. target bit) in the sources :)
While I'm at it :) what would it take to get Jeff's Rio Receiver audio device patches applied?
Thanks!
On Friday, March 29, 2002, at 12:20 AM, Tripp Lilley wrote:
In many places, the configure scripts do tests like this:
case "$host" in ...
Should those be changed to:
case "$target" in ...
?
No; I think the patch you mention calls configure incorrectly.
When cross-compiling a program, there are three potential systems involved, but usually only two. They are:
build the system which is building the program host the system which will run the program target the system which will run programs created by the program
The target system is only used in the case of building cross-compilers. Since MAD is not a cross-compiler, only the build and host systems are relevant. It's not usually necessary to specify the build system, so only the host needs to be specified if it's not the same as the build system.
You can specify the host directly as an argument to configure:
./configure arm-linux
While I'm at it :) what would it take to get Jeff's Rio Receiver audio device patches applied?
I'll have a look at it. It looks similar enough to the empeg audio module that I might be able to merge the two.
Cheers,
-- Rob Leslie rob@mars.org
-
Rob Leslie -
Tripp Lilley