Re: Repost: lock/unlock filechannel Win32 implementation (patch file)

[Michele Sandri <gpointorama at gmail dot com>]

Hi Mohan,

Pedantic is good, thank you for listing what changes i need to make :)
Reposting, hoping that this time the patch follow all the rules and 
isn't malformed!

regards,
Michele

Mohan Embar wrote:
> Hi Michele,
>
> (Be sure to CC java-patches too so we have a record
> of everything.)
>
> > I hope the layout now is ok.
> >
> > The test case is really simple, but i've tested with 
> > http://freenetproject.org/ that uses Sleepycat (Berkeley db for java).
>
> Thanks for resubmitting this.
>
> Could you make four small changes to this patch? The
> reason I'm being so pedantic about this is because I
> also like it when people are like that with me.
>
> - For the ChangeLog, I believe the correct way of saying
>   that you've implemented a new method looks like this (taken
>   from an actual ChangeLog entry):
>
> 	* gnu/awt/xlib/XToolkit.java
> 	(getLocalGraphicsEnvironment): Implemented. 
>
> - Could you keep the return value and method name on separate
>   lines, as per the other methods in natFileChannelWin32.cc?
>   This is due to some arcane GNU convention which we don't necessarily
>   respect in the rest of the Win32 code, but I think it's nice
>   to keep conventions consistent within a single source file.
>
> - For the same reasons, can you decrease the indent size from four
>   to two spaces for consistency with the rest of the file?
>
> - Finally, could you eliminate the bounds checks (and therefore
>   the import of IllegalArgumentException) from the code? As far
>   as I can see, these aren't necessary and aren't present in the
>   POSIX version.
>
> For some reason, I wasn't able to apply the patch (it complained
> about a malformed line) and had to apply it manually. Not quite
> sure why this was.
>
> I ran the test against 4.2 and it prints the expected output. I'm
> not terribly impressed with the test, but I think the patch looks
> fine and with the above changes, can commit this to 4.2 and the
> trunk. Thanks again for doing this!
>
> -- Mohan
> http://www.thisiscool.com/
> http://www.animalsong.org/
2007-01-28  Michele Sandri  <gpointorama@gmail.com>

	* gnu/java/nio/channels/natFileChannelWin32.cc
	(lock): Implemented.
	(unlock): Implemented.

Index: gnu/java/nio/channels/natFileChannelWin32.cc
===================================================================
--- gnu/java/nio/channels/natFileChannelWin32.cc	(revision 117867)
+++ gnu/java/nio/channels/natFileChannelWin32.cc	(working copy)
@@ -343,18 +343,52 @@
 }
 
 jboolean
-FileChannelImpl::lock
-(jlong /*pos*/, jlong /*len*/, jboolean /*shared*/, jboolean /*wait*/)
+FileChannelImpl::lock (jlong pos, jlong len, jboolean shared, jboolean wait)
 {
-  throw new IOException (JvNewStringLatin1
-    ("FileChannel.lock() not implemented"));
+  DWORD flags = 0;
+  OVERLAPPED ovlpd;
+
+  ZeroMemory(&ovlpd,sizeof(OVERLAPPED));
+
+  if(!shared)
+    flags |= LOCKFILE_EXCLUSIVE_LOCK;
+  if(!wait)
+    flags |= LOCKFILE_FAIL_IMMEDIATELY;
+
+  ovlpd.Offset = (DWORD)pos;
+  ovlpd.OffsetHigh = pos>>32;
+
+  DWORD lenlow = (DWORD)len;
+  DWORD lenhigh = len>>32;
+
+  BOOL ret = LockFileEx((HANDLE)fd,flags,0,lenlow,lenhigh,&ovlpd);
+
+  if(ret==ERROR_IO_PENDING && !shared && wait)
+    ret = GetOverlappedResult((HANDLE)fd,&ovlpd,NULL,wait);
+
+  if(!ret)
+    _Jv_ThrowIOException(GetLastError());
+
+  return true;
 }
 
 void
-FileChannelImpl::unlock (jlong /*pos*/, jlong /*len*/)
+FileChannelImpl::unlock (jlong pos, jlong len)
 {
-  throw new IOException (JvNewStringLatin1
-    ("FileChannel.unlock() not implemented"));
+  OVERLAPPED ovlpd;
+
+  ZeroMemory(&ovlpd,sizeof(OVERLAPPED));
+
+  ovlpd.Offset = (DWORD)pos;
+  ovlpd.OffsetHigh = pos>>32;
+
+  DWORD lenlow = (DWORD)len;
+  DWORD lenhigh = len>>32;
+
+  BOOL ret = UnlockFileEx((HANDLE)fd,0,lenlow,lenhigh,&ovlpd);
+
+  if(!ret)
+    _Jv_ThrowIOException(GetLastError());
 }
 
 java::nio::MappedByteBuffer *