Laurent Raynal

Maybe you already know about this but did you see that there is a classpath attribute you have to fill out in the forms related to the managed server in the admin console ? When you don't set up anything the classpath only contains /lib/weblogic_sp.jar and /lib/weblogic.jar.

Hello,
I'm making inquiries about a possible solution to access dBase files residing on a SCO platform from a remote Java application (J2EE server running on a UNIX machine). We obviously need some kind of server software on the SCO machine to allow JDBC access from the remote Java application.
I have an idea of the solution if the files were on a Windows machine : RmiJdbc and a JDBC/ODBC Bridge. But SCO is something different ...
Thanks

Hi,
Here is yet another ejbStore() call problem :
I have a CMP entity bean mapping a table. In my program it is created from a "wrapper" Session Bean that encapsulates data access. The session bean method has a TRANSACTION_READ_COMMITTED isolation setting. Then the method
goes on to create a related record
in another table that has a foreign key constraint to the "Entity" table,
using other classes (the Data Access Object Pattern).
The method fails throwing an SQLException saying the foreign key constraint is violated : in fact the Entity ejbStore() method is never called.
Actually, all my EJB methods (for session and entity beans) have a
TRANSACTION_READ_COMMITTED isolation setting because I previously ran into
an Oracle Serialization problem (the well-known ORA-08177 widely discussed
in other forums).
Generally, we use DAO classes to perform all "dependent" tables management,
the main table being mapped to an entity bean.
What transaction setting or strategy should I use to make this work ?
I'm using Weblo 5.1 SP 11 on NT
Thanks in advance.

Hi,
My project applied about the same architecture as Sun's Petstore 1.1.2 app. We use a statefull session bean (SFSB) as a "Client Controller" to maintain state.
Our problem is that the SessionContext principal isn't maintained between HTTP requests : it is lost on the second request, the first one after creating the SFSB.
Of course, my servlet is unauthenticated, all our authentication is done on the SFSB. It seems that web unauthentication is in some way "transmitted" to the EJB tier. Am I right ? I'm using WL 5.1 .
Also another question : I'm a bit confused about what to keep on the web tier in order to have a handle on the statefull. In Petstore, the reference kept in the HTTPSession is the EJB remote reference (in fact just the return of the create() method). It's not a javax.ejb.Handle, although it seems that it is the recommended way to do this.
Thanks a lot.

Hi,
I'm a bit confused about what to keep on the web tier to have a handle on the statefull session bean (SFSB) "Client Controller". I use Sun's Petstore 1.1.2 app as a tutorial, and what I see is that the reference kept in the HTTPSession is the EJB's remote reference (simply the return of the create() method).
I applied the same architecture in my app : I use a SFSB as a client controller to maintain state. However, I noticed that the SessionContext principal isn't maintained between HTTP requests (it is lost on the second request, the first one after creating the SFSB). My servlet is unauthenticated, all authentication is done on the SFSB. It seems that web unauthentication is transmitted to the EJB tier. Am I right ? I'm using WL 5.1 .
Thanks

Originally posted by Nitin Shivaram:
Hi Bill,
Holding a reference to the stateful bean would not be the done by the browser but by a server side application which runs in a browser, typically a servlet or a jsp.
This is how you could do it.
The reference to the stateful session bean has to be converted to an javax.ejb.Handle, to make sure you abstract the non-bean characteristics, and then store this handle in a session.
And then when you have to use it, retrieve from the session, call getEJBObject() to get the reference to the stateful bean.

Hi,
We would like to have all the (anonymous) users of a web app authenticated with the same principal (the web app "owner") to the EJB tier. Our problem is that the fact that web users are unauthenticated seem to propagate to the EJB tier. We're using Weblogic 5.1.
Users are serviced by a session statefull EJB. We set up the EJB when the user first accesses the application : user and password are set in the JNDI context and subsequent method calls on the EJB corresponding to the same HTTP request are authenticated. However, when the servlet accesses its session EJB on the 2nd hit, it won't be authenticated.
Any ideas ?

Hi,
I have an applet connecting to a PHP script that generates a data file in gzip format and then redirects to it. However, Netscape 6 doesn't seem to allow the redirect whereas Netscape 4.7x and IE 5x do. Can you confirm this ? I changed the script to output gzip data directly but I'd like to know why only N6 behaves like that (because of the Sun JVM?).
java.security.AccessControlException: access denied (java.util.PropertyPermission http.strictPostRedirect read)
at java.security.AccessControlContext.checkPermission(Unknown Source)
at java.security.AccessController.checkPermission(Unknown Source)
at java.lang.SecurityManager.checkPermission(Unknown Source)
at java.lang.SecurityManager.checkPropertyAccess(Unknown Source)
at java.lang.System.getProperty(Unknown Source)
at myClass.getURLReader(myClass.java:325)

[This message has been edited by Laurent Raynal (edited October 16, 2001).]

Hi,
I can't come up with a reliable code for reading from an URL connection on a PHP script with an applet running on Netscape 4.7. My code works on other browsers (IE 5, Netscape 6), but not consistently on Netscape 4.7 when my data (which is text/plain) is large : when I call the read() method on my InputStreamReader, Netscape4.7 won't read all the bytes whereas the buffer is large enough. I suspect my PHP script doesn't put the right headers because conn.getContentLength() for example returns -1.
The funny thing is that when I compress my data in the php script (serving a gzip file), the applet reads the output ok using a GZIPInputStream !
My code looks like this :
con = myURL.openConnection();
con.setDoOutput(true);
con.setDoInput(true);
con.setUseCaches(false);
con.setRequestProperty("Content-Type", "application/x-www-form-urlencoded");
con.setRequestProperty("Content-Length", " " + paramStr.length());
DataOutputStream out = new DataOutputStream(con.getOutputStream());
out.writeBytes(paramStr);
out.flush();
out.close();
InputStream in = new DataInputStream(con.getInputStream());
InputStreamReader urlReader = new InputStreamReader(in);
cbuf = new char[large_enough_int];
mylength = urlReader.read(cbuf);
As a result, mylength is less than the number of bytes really available from the ressource !
Many thanks in advance.

Hi all,
I want to handle null values according to the XML Schema Spec, i.e. using the following definition :
<xsd:element name="MYDATE" type="xsd:date" nillable="true" />
and then in the XML doc :
<MYDATE xsi:nil="true"/>
To validate my document, I'm using XML Spy 4.0 beta 2 which uses MSXML Parser (I installed v4.0) I guess.
My problem is that I get an error message : "Invalid value for date in element MYDATE". Checking the MSXML Parser doc I found that sometimes "nillable" appears as "nullable" and "nil" as "nill".
Is there a problem with MS implementation here ? Should I try another validating parser ? I don't think I have namespace problem in my shema.

Hi,
Can anyone recommend a components library for charting? I'm doing research for a project with mostly stock and financial charts. We need both server and client-side components : we would like to deliver charts with applets and servlets, while mutualizing the customization effort. We also want our servlets to generate charts for various platforms (wap, pda).
It seems that Sitraka and VisualMining are the leaders on the market. Any comments?
Thanks.

Thanks.
Our problem is that we don't want to limit the protocol to HTTP. For a few customers, we would like to set up a socket connection and use a faster protocol to deliver large amounts of data.

Hi all,
I'm a developer for a market data redistributor in France. I'm doing research for a project that is to provide our customers with market data (quotes, history, order books, news, etc.) using an XML-based protocol. The data is provided by our backend on which the "XML server" would interface.
That server would also be used internally by our trade applications.
Our concern is to come up with a protocol that would be easy to integrate into specific customer client applications. The best way to do this would be to implement a standard, but they seem either too large (FIX) or unfinished (MDML).
On the other hand, we would like to keep it simple to
minimize the work on our internal client applications. That would lead us to design our own protocol. We could provide our customers later with an API on it.
Which direction (standard or custom protocol) would you recommend ?
Thanks