Expose a `try_close()` (or equivalent) method for File

[andreascasapu]

NOTE: The text below is copied from this issue that was accidentally opened in hdfs-sys. It was intended for hdrs.

The docs specify:

File will hold the underlying pointer to hdfsFile.

The internal file will be closed while Drop, so their is no need to close it manually.

However, I have run into issues with the library, as sometimes closing files fails, but nothing is reported back into the Rust code. Instead, the logs show something like

[[REDACTED]] WARN hdfs.DataStreamer: DataStreamer Exception
org.apache.hadoop.ipc.RemoteException(java.io.IOException): File [[REDACTED]] could only be written to 0 of the 1 minReplication nodes. There are [[REDACTED]] datanode(s) running and no node(s) are excluded in this operation.
	at org.apache.hadoop.hdfs.server.blockmanagement.BlockManager.chooseTarget4NewBlock(BlockManager.java:2121)
	at org.apache.hadoop.hdfs.server.namenode.FSDirWriteFileOp.chooseTargetForNewBlock(FSDirWriteFileOp.java:286)
	at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getAdditionalBlock(FSNamesystem.java:2706)
[...]

Hence, I would suggest exposing a try_close() method (or equivalent) to allow users to manually close the file and handle the errors. I was thinking something like

impl File {
    fn try_close(self) -> Result<()> {
        unsafe {
            let error_code = hdfsCloseFile(self.fs, self.f);
            // hdfsCloseFile will free self.f no matter success or failed.
            self.f = ptr::null_mut();
            if error_code != 0 {
                return Err(Error::last_os_error());
            }
        }
        Ok(())
    }
}

I can give it a shot to implement myself, if you take contributions.

[andreascasapu]

MR for this is here