There is a test for Kernel32
(com.sun.jna.examples.win32.Kernel32Test) that checks for
GetLastError results; you need to make sure that GetLastError is
called once before you actually need it, so that the system calls
required to set up the calling information don't overwrite the result
you're trying to obtain.
It's possible that the VM itself has some effect on GetLastError when
making JNI calls, but I don't know of anything specific.
You might try (in your simplified JNI) to determine when the error
code gets overwritten. Any Java code that executes in between your
failed w32 call and GetLastError has a possibility of invoking some
native code that wipes the GetLastError result (I don't recall
whether the result is thread-specific or not).
On Aug 3, 2007, at 7:11 AM, Kim Valentine wrote:
Hi,
I am trying to use various Kernel32 APIs that signal errors by
returning zero or false,
but the reason for the failure has to be obtained by calling
GetLastError, I mapped that API, but always get zero returned. I
suspect that some call required by JNA is calling a windows
function that overwrites the lastError. I tried setting up my own
JNI dll that only contained a call to GetLastError, but this too
always returns zero. Has anyone any ideas how I can get this error
code?
Rgds
Kim
Yahoo! Answers - Get better answers from someone who knows. Try it
now.
1 message in net.java.dev.jna.usersRe: [jna-users] GetLastError always r...
