We Keep Coding

Java crashes when closing WDropTargetContextPeerFileStream with heap corruption

I am currently working on implementing Drag & Drop from Outlook to Swing (on Windows) using a Swing DropTarget. Because Outlook Drag and Drop does no automatically work with Swing, I debugged it and found out it used the FileNameW native for the event. To support this I use this code:
private static final String nativeFileNameW = "FileNameW";
private static final DataFlavor fileNameWFlavor = new DataFlavor(InputStream.class, nativeFileNameW);
public void installFileNameWFlavorIfWindows(DropTarget dt) {
FlavorMap fm = dt.getFlavorMap();
if (!(fm instanceof SystemFlavorMap)) {
fm = SystemFlavorMap.getDefaultFlavorMap();
}
if (fm instanceof SystemFlavorMap) {
SystemFlavorMap sysFM = (SystemFlavorMap) fm;
sysFM.addFlavorForUnencodedNative(nativeFileNameW, fileNameWFlavor);
sysFM.addUnencodedNativeForFlavor(fileNameWFlavor, nativeFileNameW);
dt.setFlavorMap(sysFM);
}
}
It seems to work fine, but I am not sure if this is the correct approach, since I couldn't find any resources on this problem.
In the drop event I can now get an InputStream when an Outlook Email is dropped on the Swing Component. I use the following code in my drop method (the real method is more complex, because it also handles other DataFlavors, but this example here can reproduce the error):
public void drop(DropTargetDropEvent dtde) {
Transferable transfer = dtde.getTransferable();
boolean accepted = false;
if (transfer.isDataFlavorSupported(fileNameWFlavor)) {
accepted = true;
dtde.acceptDrop(DnDConstants.ACTION_COPY);
try (InputStream is = (InputStream) transfer.getTransferData(fileNameWFlavor)) {
//Do something with InputStream
} catch (UnsupportedFlavorException e) {
e.printStackTrace();
} catch (IOException e) {
e.printStackTrace();
}
}
dtde.dropComplete(accepted);
}
I use a try with resource statement to ensure the stream is closed after the drop event. I want to close the stream to make sure there are no open File Handles or similar native resources, that could be limited, after the drop is completed.
The InputStream for a Drop from Outlook is an Instance of WDropTargetContextPeerFileStream and when the close method is called, it crashes in the native Method freeStgMedium, which should free the native windows data structure.
I do not get any error output on the command line.
The Program terminates with error code -1073740940 which seems to indicate a heap corruption error.
Is there anything I am missing? Is this InputStream not supposed to be closed or is there a Bug earlier on.
I am using the JDK from Azul, Zulu 8.48.0.53 (Java 8u265).
I have also tried it with Zulu 11, Oracle Java 8 and a Redhat build of Openjdk 8, all fail the same way.
Update:
I think I tracked the bug down to JDK native code, that gets the data.
The JDK Code creates a STGMEDIUM object on the stack and passes a Pointer to that to the Windows Method IDataObject::GetData(). This method writes its data back into the STGMEDIUM* parameter.
This should not be a problem, since all examples of this Windows function did it the same way. But it seems, that Outlook does not initialize the member variable IUnknown *STGMEDIUM::pUnkForRelease, but instead relies on the caller to zero-fill the data structure (or Outlook has a Bug).
When the native resources are released by Java, it calls ReleaseStgMedium, which tries to call Release on the pUnkForRelease pointer, if it isn't NULL, which causes the error.
For now, I simply don't close the input stream and let a FileHandle leak, which is not optimal, but I don't see any other solution.
If I find a real solution to this Bug, I will write an Update/Answer here.

JAVA 10 - java.sql.SQLException: No suitable driver found for jdbc:mysql://localhost:3306/ error

I know it's an old question but the case is different here. It seems to be a bug at JAVA 10 or MySQLConnector/J 8.0.12 jar classes.
The same projects (lots of them, not just one) that worked fine with JAVA 8 and MySQLConnector/J 5.1.39, which I used before, now don't work after I updated my system.
But the error appears only with RowSet. With DriverManager/Connection/ResultSet, explicitly used, it works pretty fine with the same url, url-parameters and MySQL driver.
I've tried to redundantly use Class.forName("")... as well as variations of the url parameters, but nothing works when I use RowSet. The same java.sql.SQLException: No suitable driver found for jdbc:mysql://localhost:3306/scheme1?verifyServerCertificate=false&useSSL=true is thrown again and again.
Why the same code works with Java 8 and not Java 10 ?
try {
RowSet rowSet = RowSetProvider.newFactory().createCachedRowSet();
String url = "jdbc:mysql://localhost:3306/scheme1?verifyServerCertificate=false&useSSL=true";
rowSet.setUrl(url);
rowSet.setUsername("root");
rowSet.setPassword("root");
rowSet.setCommand("select * from client");
rowSet.execute();
while(rowSet.next()) {
System.out.println(rowSet.getInt(1)+" "+rowSet.getString(2));
}
} catch (SQLException e) {
e.printStackTrace();
}enter image description here
(See pictures with the error).
If someone know the problem and a solution please help. I'd like to keep using Java 10.
Daniel Pinheiro
danielpm1982#gmail.com

mysql-connector.jar is ment for JAVA 8
You need to wait till MySQL rollout the connector for JAVA 10 or downgrade to JAVA 8.

how windows will do auto java update checks?

First of all sorry for my poor English.
I want to know how the windows will do auto java update check behind the User Interface ?
The UI will just react based on our input which is in the link , http://java.com/en/download/help/java_update.xml#howto .
But , how windows checks the updates programmatically.
I wrote a small program in java ,
public class JavaLatestVersion {
public static void main(String[] args){
try {
BufferedReader br = new BufferedReader(new InputStreamReader(new URL(
"http://java.com/applet/JreCurrentVersion2.txt").openStream())) ;
String fullVersion = br.readLine();
System.out.println("fullVersion : "+fullVersion);
String version = fullVersion.split("_")[0];
String revision = fullVersion.split("_")[1];
System.out.println("Version " + version + " revision " + revision);
} catch (IOException e) {
e.printStackTrace();
}
}
}
My questions :
1. Is the above program is the reliable way to get the latest java version ? Or any other standard way to get the latest java version (Not in the computer) ?
2. Is windows use the same way to determine the latest java version ?
3. Is windows use this link for updates http://java.com/applet/JreCurrentVersion2.txt ?
Any one know the secret code behind how windows will check for latest java updates?
Thanks in advance.

Checking for Java updates is done by Java Auto Updater. It is ordinary application (which is ran when Windows starts up).
Yes, it is reliable way to get the latest version of Java (as updater can not only update Java, it can also update itself). But pay attention to firewall/group policy settings which can prohibit updater to access the Web.
Windows doesn't update Java.
Only debugging Java Auto Updater can help to determine what URL it uses.
Unfortunately, Java Auto Updater has only graphical interface and hides all work behind the scenes. So finding a "secret code" is not easy to do. All the more in many cases reverse-engineering non-open source software is illegal from a license point of view.
URL that you provided above doesn't works. Because it says 8.0_51. But latest version of Java on Downloads page is 8u65 / 8u66.
Seems that latest available version (as plain text) can be determined only by fetching http://www.oracle.com/technetwork/java/javase/downloads/index.html web page, then parsing it, handling cases when page is moved to another location, etc.

Data in J2ME RecordStore does not persist across sessions

I'm building a mobile app with J2ME, and I've found that the data I write into a RecordStore can be accessed while the program is still running but it is lost after quitting and restarting it. No exception is thrown, the data is simply lost.
UPDATE: Thanks everyone for your suggestions. I'm using NetBeans on Windows 7. I'm not sure if it is using the WTK version I have previously installed or another one it has installed somewhere else. I've checked my WTK folder for the files Pavel wrote about, but couldn't find them. Now I'm testing the features requiring persistence on my phone and everything else in the emulator, but it would of course be much better to be able to test everything in the emulator.
private RecordStore recordStore = null;
public MyMIDlet() {
readStuff(); // output: nothing found in recordStore :(
saveStuff();
readStuff(); // output: stuff
}
private void readStuff() {
try {
recordStore = RecordStore.openRecordStore(REC_STORE, true);
int n = recordStore.getNumRecords();
String stuff;
if (n == 0) {
stuff = "nothing found in recordStore :(";
}
else {
stuff = new String(recordStore.getRecord(1));
}
System.out.println(stuff);
}
catch (Exception e) {
System.out.println("Exception occured in readStuff: " + e.getMessage());
}
finally {
if (recordStore != null) {
try {
recordStore.closeRecordStore();
}
catch (Exception e) {
// ignore
}
}
}
}
private void saveStuff() {
try {
recordStore = RecordStore.openRecordStore(REC_STORE, true);
int n = recordStore.getNumRecords();
byte[] stuff = "stuff".getBytes();
recordStore.addRecord(stuff, 0, stuff.length);
} catch (Exception e) {
System.out.println("Exception occured in saveStuff: " + e.getMessage());
} finally {
if (recordStore != null) {
try {
recordStore.closeRecordStore();
} catch (Exception e) {
// ignore
}
}
}
}

If you use Sun WTK, it creates a file named "in.use" in its "appdb" folder:
C:\WTK25\appdb\DefaultColorPhone\in.use
If you close your emulator in unusual way (kill a process, for example), it would not delete it, and next time you run emulator, it would create temporary folder for storing data:
C:\WTK25\appdb\temp.DefaultColorPhone1
when starting this way, it should print in console: "Running with storage root temp.DefaultColorPhone1".
I fix it, including into my ".bat" file a line for deleting "in.use" file each time, emulator runs. But you should be careful when running several emulators at once.

I experienced the same problem myself, I did however discover that NetBeans, or whatever, deletes the deployed program files after execution. These files are located in the C:\Documents and Settings\MyUser\javame-sdk\3.0\work\0\appdb folder, might be different on Vista/Win7 and I guess the number in the path refers to the emulator you are currently using. Anyways, in this folder look for something that is named like your RecordStore. E.g. "00000002_PSC_onfig.db", which is my suite configuration recordstore named PSConfig. By copying this to e.g. "Copy of 00000002_PSC_onfig.db" it will not be deleted. After NetBeans have cleaned up, just copy it back to its original name.
The next time you hit run in NetBeans your recordstore will be there. It's pain, but at least it gives you the possibility to use the emulator to debug your RMS handling.

This question has been around for a while but I stumbled upon it whilst looking for an answer to the same problem with the emulator but in my case it was when using the Java ME 3 SDK. It is possible that the solution I found might also fix this problem.
Using:
emulator -Xdescriptor:/path/to/app.jad
will according to the docs: "Install a MIDlet, run it, and uninstall it after it finishes."
To persist an installation (and it's data) you should use:
emulator -Xjam:install=<JAD-file-URL>
The JAD file URL can either be a web address or 'file:///path/to/app.jad' if you want to install from your local file system. This installation command will display an application storage number which you can then use to launch the emulator and run the previously installed app by calling:
emulator -Xjam:run=<application-storage-number>
See the docs for further command line options.

I could finally get it to work on a real handset. It seems that, as Martin Clayton suggested, the emulator reset erased the data. I'm still looking for a way to enable persistence in the emulator though.

If you are using windows Vista there can and almost are permission issues. I am not sure how to resolve this but you might want to check that the user that is running the emulator has access to write to the emulator store.
In appdb/$phone/*.db

to fix the storage problem you need to check the option "Storage size" in the netbeans platform manager.
Go in project properties;
go in platform;
manage emulators;
select the sun java wireless toolkit;
go in Tools and Extensions;
open preferences;
storage;
storage size option... set a size.
this works for me

I experienced the same issue on Ubuntu Linux and working with WTK 2.5.2 and Netbeans 8.0.2. I later figured it was caused by shutting down my laptop without closing the emulator. It also happens if you run a second emulator without shutting down the first.
My solution is based on the Best Answer here, just shut down all emulators and delete the file located at
~/j2mewtk/2.5.2/appdb/DefaultColorPhone

How can I use the eclipse indenter from my code?

I noticed the eclipse indenter has support for the latest version of java, and it would be nice if I could use that class to indent generated java source code. Is there a way of integrating it ?
EDIT: I need to be able to include the code formatter in my code. No external calls.
EDIT2: I've managed to get it working. You can read the story here. Thanks VonC !

You can try running the formatter as a standalone application (also detailed here).
eclipse -vm <path to virtual machine> -application org.eclipse.jdt.core.JavaCodeFormatter [ OPTIONS ] <files>
Try first to define formatting settings with eclipse IDE in order to achieve the right result, then export those settings, and use that configuration file in the eclipse.exe parameters.
Or see also "Generating a Config File for the Formatter Application"
eclipse [...] -config <myExportedSettings>
In a java program, you can try to directly format by:
Creating an instance of CodeFormatter
Using the method void format(aString) on this instance to format aString. It will return the formatted string.
Thanks to Geo himself and his report in his blog entry, I now know you need to use DefaultCodeFormatter
String code = "public class geo{public static void main(String[] args){System.out.println(\"geo\");}}";
CodeFormatter cf = new DefaultCodeFormatter();
TextEdit te = cf.format(CodeFormatter.K_UNKNOWN, code, 0,code.length(),0,null);
IDocument dc = new Document(code);
try {
te.apply(dc);
System.out.println(dc.get());
} catch (MalformedTreeException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (BadLocationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
Again, full details in the blog entry. Thank you Geo for that feedback!
Thorbjørn Ravn Andersen mentions in the comments:
Maven2 Java Formatter Plugin v0.4 describes a maven plugin that allows Maven to invoke the Eclipse formatter.
As of 0.4 it invokes Eclipse 3.5 which does not support Java 8.

Actually, there is one problem with VonC's answer: DefaultCodeFormatter is in an 'internal' package, and therefore should not be be used by clients!
I recently asked the same question here on stackoverflow, and came up with the answer a little while later.
In short, you need to use ToolFactory, as in
ToolFactory.createCodeFormatter(null);

I was using the CodeFormatter in an Eclipse-independent project. The default options used by calling ToolFactory.createCodeFormatter(null); could not handle the source code - the result of the format() call being null.
A minimal working options setup is the following:
Hashtable<String, String> options = new Hashtable<>();
options.put("org.eclipse.jdt.core.compiler.codegen.targetPlatform", "1.8");
options.put("org.eclipse.jdt.core.compiler.compliance", "1.8");
options.put("org.eclipse.jdt.core.compiler.source", "1.8");
CodeFormatter codeFormatter = ToolFactory.createCodeFormatter(options);