Ah, you caught me typoing. The correct path to the unfound include file is
indeed
../jdk1.6.0._01/include/win32/
The JDK I'm using is Sun's.
Looking at the args sent to gcc, one of the very curious -I settings is
...jdk1.6.0_01/include/MINGW32_NT-5.1
The problem stems from MinGW's uname returning this:
MINGW32_NT-5.1
So near the top of jna-source/jnalib/native/Makefile, the sed command needs
another case
-e 's/MINGW32.*/win32/g' \
and all should be ducky.
But then I got past that, ran into other problems (missing
com_sun_jna_Function.h), which
makes me think I should have run ant. So, off to do that, and it seems that
while running
ant under cygwin might work, running ant under MinGW seems to be rough sledding.
So at this point, I'm abandoning MinGW and going with cygwin...
---- Timothy Wall <twal...@dev.java.net> wrote:
What JDK are you using? Sun's JDK for windows puts jni_md.h into
$JDK_HOME/include/win32, not 'w32'. We can always tweak the makefile
to support variants...
T.
On Jul 9, 2007, at 11:45 AM, <bb...@cox.net> <bb...@cox.net> wrote:
Seeing the recent changes since the last jna.jar available for
download was built, I've set out to build my own.
I'm trying to build on WinXP. I decided to use MinGW and MSYS as
build tools for the native stuff. Using "make CC=gcc", it fails
trying to compile the first file, distpatch.c.
The reason is that the include path seems to be not quite right.
It can find jni.h in .../jdk.6.0._01/include/
But jni.h includes jni_md.h, and that is not found, because it is in
.../jdk.6.0._01/include/w32/
I can guess how to fix this by hand, but suspect I am doing
something wrong overall...
3 messages in net.java.dev.jna.usersRe: [jna-users] MinGW build can't fin...
