Different behavior in JWS vs. JDK environments

Hi Folks,
I've got an app that I'd successfully tested on a number of systems that had a full JDK install with the JWS "update". But, my customer reported an odd error I couldn't duplicate. But by setting up a box with just the JWS "full package" (no JDK install) and running my app via JWS on that box, I've duplicated the problem. So, it seems like something's different about the run-time environments in the JWS vs. the JDK set-ups.
On the JWS-only box, my JDBC driver is reporting a character conversion exception (unsupported encoding--which is totally bogus: it's a very commonly-used encoding for Sybase DB's):
java.sql.SQLException: JZ006: Caught IOException: java.io.IOException: JZ0I6: An error occured converting UNICODE to the charset used by the server. Error message: java.io.CharConversionException: java.io.UnsupportedEncodingException: Cp850
     at com.sybase.jdbc2.jdbc.ErrorMessage.raiseError(ErrorMessage.java:487)
     at com.sybase.jdbc2.jdbc.ErrorMessage.raiseErrorCheckDead(ErrorMessage.java:723)
     at com.sybase.jdbc2.tds.Tds.handleIOE(Tds.java:3071)
     at com.sybase.jdbc2.tds.Tds.language(Tds.java:702)
     at com.sybase.jdbc2.jdbc.SybStatement.sendQuery(SybStatement.java:1424)
     at com.sybase.jdbc2.jdbc.SybStatement.execute(SybStatement.java:1662)
     at com.sybase.jdbc2.jdbc.SybStatement.execute(SybStatement.java:773)
... back down into my application.
Now, since the application works fine on the boxes I've got with the full JDK, can anyone suggest what might need to be shipped along with this app to allow these char conversions to happen properly? I'm assuming the JDBC driver is trying to call a conversion routine in the JRE that either has different limitations in the JWS-only environment or something of the like. (Irritatingly, since it's the JDBC driver throwing the exception, I can't even see what it's calling). One other difference that's just occurred to me: the JDK installs are all at 1.3.1 level (and I think the JWS is a slightly different JRE--I have the latest thing from the JWS download page).
Any suggestions welcomed! (I'm sure hoping to avoid telling my customers they have to download the JDK.)
John

Turns out the answer is that even what seem like pretty run-of-the-mill character encodings aren't supported by the Windows JRE that's included with the "Full Package US-English Only" install of JWS.
When I switched out the JWS-installed JRE for the one that comes with the "International" version, everything worked again. Here's a link that describes how the Windows JRE's are different from those on other platforms (Windows JRE's come in two flavors--those that support various encodings and those that support only a select few):
http://java.sun.com/j2se/1.3/docs/guide/intl/encoding.doc.html
CFN,
John

Similar Messages

Different behavior for index

Hi.
I have the table inl_allocations with following index:
CREATE INDEX inl_allocations_n1 ON inl_allocations
( ship_header_id                  ASC,
    adjustment_num                  ASC,
    ship_line_id                    ASC )
When running query below, we have different behaviors for indexes in different DB:
SELECT *
FROM INL_ALLOCATIONS CA, INL_SHIP_LINES_ALL CC
WHERE CC.SHIP_LINE_ID = CA.SHIP_LINE_ID
AND CC.SHIP_HEADER_ID = CA.SHIP_HEADER_ID
1- For DB 1 (11.2.0.1.0) the explain plan for query:
                (3) INDEX INDEX FULL SCAN INL.INL_ALLOCATIONS_N1 [Analyzed]
                    Est. Rows: 33 Cost: 1
2- For DB2 (11.2.0.3.0) the explain plan for query does not use the index INL_ALLOCATIONS_N1, and we have a FULL in table.
           (3) TABLE TABLE ACCESS FULL INL.INL_ALLOCATIONS [Analyzed]
           (3)   Blocks: 13 Est. Rows: 167 of 269 Cost: 5
We would like to know the reason why there are differences between both environments.
Thanks and Regards.

many years ago Oracle used to optimize queries with a set of rules that ranked the potential access paths and finally chosed the possible operation with the highest rank. So for two systems with the same DDL Oracle would have generated the same plan. This was the rule based optimizer (RBO) and its shady ghost still dwells somewhere deep in the Oracle code. Since at least Oracle 8i (> 10 years ago) the standard optimization engine is the cost based optimizer (CBO) which uses statistics to generate execution plans: and if the data of your systems is different (or even if the data is identic but its physical order is different and changes the index clustering factor) the CBO can and will generate different plans and use or not use existing indexes.
Regards
Martin

ESS Benefits Generic Enrollment - Different behavior backend v/s portal

We are using the ESS Benefits generic enrollment service for enrollment kicking off the adjustment reasons on IT 0378.
Here is the configuration set up, trying to simplify to explain the matter. We are on ECC 6.0 with EHP2.
1. Configured one plan type in savings category.
2. Within this plan type, configured 2 plans.
3. Configured all other related configuration to set up the 2 plans.
4. Configured the adjustment reason and the appropriate plan type created in step 2.
5. Created the required adjustment reason on 378.
In HRBEN0001 transaction via backendwe can see the 2 plans configured within the plan type.
But when Trying to enroll from ESS - only one plan can be seen for that plan type, the other plan is not shown at all.
OK - eventually, debugged and found this piece of code in FM HR_BEN_ESS_FILL_SELECTION_DISP. The 2 different plans are there until this piece of code takes one of them Off...
CODE -
3.) Reduce entries to one entry per type and period
delete adjacent duplicates from selection_display_gt
    comparing pernr
              barea
              bpcat
              pltyp
              begda
              sprps.
CODE -
This FM is called only for ESS application and not for backed program.
What we do not understand is why is it behaving differntly on backend v/s portal.
Any information or experience with this form any one of you will be highly appreciated.
Thanks.

Siddharth,
Thanks for your reply. Yes ESS has its own FMs. The UI, user exits, BADIs etc are different for backend v/s front end.
What we are not clear with, is the different behavior in core functionality -- i.e. When the adjustment reason is processed from the backend using HRBEN0001 - The 2 plans within the Plan Type are available for elections. BUT the same adjustment reason for the same penr when processed from ESS - there is just one plan available for elections.
It's clear from the code, that it is deleting one plan and this code is within the FM which is called only from ESS. But we are still not sure why it should behave differently i.e. 2 plans via backend v/s one plan via ESS.
Wanted to see if anyone can share any information if they have a similar benefits set up .. 2 plans within one plan type FOR savings plan category and then having a adjustment reason set up.
Regards.

Different behavior of movewindow with different windows is (xp or windows 7)

we have a windows wpf stand alone application which must be integrated within a windows form 2.0 stand alone application. given to the particular implementation of the windows form application we decided to use a light integration trying to change the parent
handle of the wpf main window setting as parent the handle of a windows of the windows form 2.0 application.
in order to reposition the wpf windows on top of the parent windows form window, we have a thread which keeps doing two operations: (1) invoke getwindowsrect on the parent windows form windows to get the parent absolute position and (2) invoke movewindow
on the wpf main window handle to reposition the window.
when we move the windows form window, the wpf windows follows the parent remaining in the correct position. we tested this solution with windows xp, windows 7, windows 8 in our office. when we went to the customer to complete the installation, both on a
windows xp and on a windows 7 the behavior was different. the size of the wpf window was much bigger than the size of the windows form parent window. back in office we found the same strange behavior in windows xp wirtual machine. in order to fix the problem
(at least partially) we had to change the operation sequence within our code as follows in the monitoring thread: (0) get parent windows position with getwindowrect (1) set the parent of the wpf main window to (null) 0 invoking setparent, (2) invoke the movewindow,
(3) re set the parent handler of the wpf main window to the handler of the windows form window.
with last solution it works but clearly there are more flikering when the window is moving or resizing respect to initial implementation.
any idea about this different behavior of movewindow function? any suggestion?
thanks
Paolo

Hi Paolo,
You'll need to post it in the dedicated WPF or Windows Forms forum for more efficient responses, where you can contact the experts.
http://social.msdn.microsoft.com/Forums/en-US/home?forum=wpf WPF forum
http://social.msdn.microsoft.com/Forums/windows/en-US/home?forum=winforms Windows Forms forum
We are trying to better understand customer views on social support experience, so your participation in this interview project would be greatly appreciated if you have time. Thanks for helping make community forums a great place.
Click
HERE to participate the survey.

Different behavior with decimal data on same version of CF

I have a linux/apache/mysql/CF MX 7 Enterprise server. I insert decimal data using the cfqueryparam that looks like
<cfqueryparam value="#BudgetSet.BUD_SET_DETAIL_COST_PER#" CFSQLType="CF_SQL_DECIMAL">.
No problems the decimal data does not get rounded, 123.45 inserts as 123.45.
I'm building applications for my neighborhood association, so I'm having godaddy.com host the web site with CF and MySQL.
Like a good developer, I took code from my CF server, moved some of the more useful programs to the godaddy.com, set up the database and started testing some apps. Now any decimal data inserted gets rounded using the same cfqueryparam syntax from my CF server (like above). The decimal data on the godaaddy.com inserts as 123.00, when I'm trying to insert 123.45, of course if I insert 123.99 I get 124.00, rounding. So I called godaddy.com and of course I get the standard technical answer "its your code". So off to testing I go. Only to determine, by not using the cfqueryparam, I could stop the rounding, more testing I determined that I need to use
<cfqueryparam value="#BudgetSet.BUD_SET_DETAIL_COST_PER#" CFSQLType="CF_SQL_MONEY"> or
<cfqueryparam value="#BudgetSet.BUD_SET_DETAIL_COST_PER#" CFSQLType="CF_SQL_DECIMAL" scale="2">. The version of CF @ godaddy.com is also MX 7 Enterprise. Finally the question, what in configuration could cause such different behavior ?
Thanks for reading my short novel.

The database drivers determine how the cfqueryparam values are
interpreted (behind the scenes ColdFusion generates a prepared
statement using the driver interface and sets the values using
standard JDBC calls). Unfortunately, like you found out, not all
drivers are created equal and have slightly different behavior or
defaults.
I would check and compare the driver versions.
Mack

Assigning different behaviors to instances of a symbol

Hello all ,I want to assign different behavior (action
script) to different instances of (a) symbole ,what should I do?I
appreciate you .Sincerely Mohsena

yes it works ,thank you very much for your attention!
sincerely yours Mohsena

What different of all version of jdk(include j2sdk)

hi everyone:
I have developed an application. But there are some other applications worked with our application. And these applications are written by java, and compiled and run by different version of jdk(or j2sdk) correspondingly. And these applications are B/S structure, the browser is IE. The situation is: if one version of jre is selected for IE, this application can run normally, but others may not run normally.
So it puzzle me.
I have 2 problems:
1. How to detect the current version of JRE(including SUN JVM and microsoft JVM) with IE?
2. What different of all version of jdk(include j2sdk)? Or where can i find the related document of the different?

I'm not sure if you want to programatically detect the VM version or just be able to see what version it is. You could go to Internet Explorer's Advanced Tab and see what version is being used. If you want to detect or change it programatically, the only way I can think of is to change the windows registry entry (you'll need to verify this, I'm not sure this is the right key) in:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Internet Explorer\AdvancedOptions\JAVA_SUN\SELECT and change the Value of RegPath to the one you want. The JVMs can be found at http://java.com/en/index.jsp. You can also search Microsoft's website for JVM information.

RDCMan different behavior for different machines

I'm using Remote Desktop Connection Manager, which has been very useful since I have to remote into so many different machines.
I am curious though and this is more of a question than a problem, but I notice different behavior remoting into different systems.
On some servers, I get this popup when I connect:
On other servers, I get brought to this screen to enter my password:
I've checked the group policy settings, registry settings, made sure I didn't have any saved credentials, makes no difference.
Any idea why?

RDP Session Setting:
Console Session Settings:
Hi,
As per my research, that is the default behavior and it correlates for 2 different scenario “Console and RDP”. When you are performing “Console” of server then it will provide you the direct screen of server to login. But when you want to take RDP
then it will ask you as “Windows Security” login prompt.
Please check my words on your screenshots
Hope it helps!
Thanks.
Dharmesh Solanki
Please remember to mark the replies as answers if they help and unmark them if they provide no help. If you have feedback for TechNet Support, contact [email protected]

Different complining result between SUN JDK 1.4 and IBM JDK 1.3.1

Hi! I simplifed my code as following:
public class TestException2 {
    public Object test() {
        java.sql.Connection cn = null;
        try {
             // 1st. close
            cn.close();
        catch (java.sql.SQLException se) {
             // 2nd. close
            cn.close();
        finally {
            return new Object();
}But I get different compiling result betwueen SUN JDK 1.4.2 and IBM JDK 1.3.1 (I have to deploied my app to the websphere 5.0)
In Sun JDK 1.4.2, I only got a warning:
[e:\work\test]javac TestException2.java
TestException2.java:14: warning: finally clause cannot complete normally
        ^
1 warningIn IBM JDK 1.3.1, I got an error:
TestException2.java:10: unreported exception java.sql.SQLException; must be caught or declared to be
thrown
            cn.close();
              ^
1 errorWhat happend? Who is right?
Thank you!
[ I believe the code should not pass the compilation ... ]

Many other people agree with you, including some Sun engineers.
Perhaps Sun just feels an obligation to continue to allow it for legacy reasons.
You might also investigate the compiler that comes with Eclipse.
http://www.eclipse.org/
It has a more flexible approach: for some cases, it lets you decide what is a warning and what is an error.

Different behavior from the PRE function based on BMRS and WMRS

Hi All,
We are seeing different behavior from the PRE function based on whether the type is Before Modified Record Saved or When Modified Record Saved. Should the PRE function work the same for both as long as there is not a Wait action in the WMRS workflow?
(((PRE('<plTier_ITAG>') <> [<plTier_ITAG>]) AND ([<plTier_ITAG>] = '2 - BA')) AND [<IndexedPick4>] = 'FSCM')
Explaination: The Tier field changes AND now equals Tier 2, AND Application is equal to 'FSCM'
The same expression fires only as designed in BRMS, but in WMRS it fires every time a SR is in Tier 2 and IndexedPick4 = 'FSCM', regardless of whether a change was made to the Tier field.

Try changing your syntax around to something like -:
PRE('<plTier_ITAG>') {NOT EQUAL} '2 - BA' AND [<plTier_ITAG>] = '2 - BA' AND [<IndexedPick4>] = 'FSCM'
+i.e. Previous value of Tier is not equal to "2 - BA" and current value of Tier equals "2 - BA" and IndexedPick4 equals "FSCM". Replace {NOT EQUAL} with the appropriate syntax+
Regards,
Cam

Different Versions of JWS Don't work with my jnlp - please help!!

I had jdk1.5.0_14 installed with IIS serving a jnlp file that indicates the java version must be 1.5. I can download and run my application from a web page.
I installed jdk1.6.0_02. JAVA_HOME and the path environment variables both point to jdk 1.5.
I try to download the same jnlp via web browser. The JWS version indicates 1.6 but the java version correctly indicates 1.5. The problem, however, is that my app no longer runs - I get an error as my app tries to install some third-party tools that only run with Java 1.5 (the app can't run with Java 1.6 but I need Java 1.6 for other apps).
When I run this command "C:\Java\JDK1.5.0_14\bin\javaws 'C:\IIS\appdirectory\myfile.jnlp'" my application runs as it did previously.
Can anyone tell me if I can tell the system which version of JWS to use when running my jnlp file? I don't want to use the Java 1.6 version for my version that will be using JDK1.5.
Help would be greatly appreciated!!!

Thanks for your reply, however, I do have the j2se version tag set and when I have a Java Console window running along with my download the first line in the console window indicates it is using JWS version as 1.6 but the second line in the window indicates it is using JRE version as 1.5...I would like for both of them to be the same. So, I am getting the correct java version but something is happening with the JWS using the version from 1.6. Here's a snippet from my jnlp file:
<resources>
     <j2se version="1.5"/>
     <jar href="./ShipControlJavaPrototype.jar" main="true" />
               <jar href="./vecmath.jar"/>
     <jar href="./xith-tk.jar"/>
               <jar href="./joal.jar"/>
               <jar href="./log4j.jar"/>
     <jar href="./lwjgl.jar"/>
     <jar href="./vorbis.jar"/>
     <jar href="./xith_utilities.jar"/>
               <jar href="./swing-layout-1.0.jar"/>
               <jar href="./sqljdbc.jar"/>
     <nativelib href="joal-win.jar"/>
     <nativelib href="jogl-natives-win32.jar"/>
               <nativelib href="joystick.jar"/>
               <nativelib href="commsutilities.jar"/>
</resources>

Bug or Feature? Different behavior after 1.6.0_02 in both Firefox & IE6/IE7

Greetings,
I have been testing a small java applet for a phpbb mod, named Proxy Revealer
Basically the java applet "phones home" (connects back to the http host serving the applet) via a Socket connection to establish a direct connection, then basically writes a custom HttpRequest string with some parameters passed to it from the php script's HTML, so it looks like:
GET /probe.php?mode=java&ip=initial_external_ip&extra=random_unique_identifier&local=LAN_IP&vendor=java_vendor&version=num&user_agent=browser
which would allow the php script on the host/site to verify the client's IP address, and compare it with the initial IP that requested the page which loaded the applet, as well as Internal LAN IP if user is behind a router/NAT
This allows the php script to unmask & log proxied users for security purposes, in case of spamming/trolling on the forum, in an attempt to thwart the average spammers/trollers at the least.
The applet and the code works as desired and in various browsers, but only on JRE 1.6.0_02 and earlier releases...
It seems ever since JRE 1.6.0_03, this has stopped working when the end-user has HTTP Proxy configured in the browser.
I tested latest JRE as of date, ver 1.6.0_07, with IE6, IE7 & Firefox 2.0.0.15 and 2.0.0.16 and it still exhibits this odd behavior.
This is what appears in Java Console when I try to visit the page serving this applet with an HTTP proxy configured in browser:
java.security.AccessControlException: access denied (java.net.SocketPermission xxx.xxx.xxx.xxx:80 connect,resolve)
xxx.xxx.xxx.xxx is the resolved IP of the server host that is hosting the applet, so basically the same origin....
If I disable the HTTP Proxy configured in the browser, the applet connects back fine. So it only happens when HTTP Proxy is configured in browser..
Is this a new feature or a bug??
Here's the java applet code:
// httpRequestor.java
// Copyright (c) MMVI TerraFrost
// Licensed under the GPL.
import java.applet.*;
import java.net.*;
public class HttpRequestor extends Applet
     public void start()
          try
               String javaVendor = System.getProperty("java.vendor");
               String javaVersion = javaVendor.startsWith("Microsoft") ? System.getProperty("java.version") : System.getProperty("java.vm.version");
               Socket sock = new Socket(getParameter("domain"), Integer.parseInt(getParameter("port")));
               String path = getParameter("path")+"&local="+sock.getLocalAddress().getHostAddress()+
                    "&vendor="+URLEncoder.encode(javaVendor, "UTF-8")+
                    "&version="+URLEncoder.encode(javaVersion, "UTF-8")+
                    "&user_agent="+URLEncoder.encode(getParameter("user_agent"), "UTF-8");
               String httpRequest = "GET "+path+" HTTP/1.0\r\nHost: "+getParameter("domain")+"\r\n\r\n";
               sock.getOutputStream().write(httpRequest.getBytes());
               sock.getInputStream();
          catch (Exception e)
               e.printStackTrace();
}and the relative portion from the probe.php script loading it:
          $java_url = $path_name . "probe.$phpEx?mode=java&ip=$client_ip&extra=$sid,$key";
?>
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<title></title>
</head>
<body>
<applet width="0" height="0" code="HttpRequestor.class" codebase=".">
<param name="domain" value="<?php echo $server_name; ?>">
<param name="port" value="<?php echo $board_config['server_port']; ?>">
<param name="path" value="<?php echo $java_url; ?>">
<param name="user_agent" value="<?php echo htmlspecialchars($HTTP_SERVER_VARS['HTTP_USER_AGENT']); ?>">
</applet>
</body>
</html>A barebone/proof-of-concept demo (also employing a couple other tricks to detect CGI proxies in the same page):
http://www.frostjedi.com/terra/scripts/ip_unmasker.php?mode=utf16
Another similar java applet code (with demo) is found towards the bottom of this page:
http://www.burghardt.pl/2008/05/web-browser-anonymity-threats/
Both of these demos work with JRE versions 1.6.0_02 and older, but fail to work (in both IE & FF) with JRE 1.0.6_03 and newer - up to 1.0.6_07 which is latest as of date
Thanks,
Jasmine

Thought I might elaborate on the details of the bug report I made, perhaps someone needs to add to it or needs to understand the problem better.
Description:
Unsigned applets cannot connect back via Socket to the originating host (from codebase) when a Proxy is configured in user's browser (IE7/IE6//FF3/FF2 tested) and when origin host's IP address doesn't resolve back to the same hostname.
Example:
www.hostingsite.com resolves to 1.2.3.4
but, 1.2.3.4 resolves back to 4.3.2.1-somewebhost.com
Affects JRE versions 1.6.0_03 - 1.6.0_07
This wasn't a problem in 1.6.0_02 or prior versions according to my tests.
An AccessControlException is thrown about SocketPermission:
I believe the security manager is doing unnecessary lookups, even after the resolved IP matches to the IP of the origin host. This is apparent from the fairly long delay before the ACE is thrown about SocketPermission.
Steps to Reproduce:
1. Configure HTTP Proxy in browser (IE/Firefox)
2. visit an html page that embeds a simple applet that tries to connect back to origin host whose hostname resolves to an IP address but the IP address resolves to a different hostname. (example code below)
Expected Result:
Applet should be able to connect back to originating host via Socket connection or write (post) to a URL on origin host (which requires a new Socket connection back)
Actual Result:
Socket sock = new Socket(Proxy.NO_PROXY);
InetSocketAddress sockAddress = new InetSocketAddress(getCodeBase().getHost(), port);
sock.connect(sockAddress);The above snippet of code would throw an ACE about SocketPermission when applet tries to initiate sock.connect
Moreover,
URL urlRequest = new URL(this.getCodeBase()+path);
HttpURLConnection conn = (HttpURLConnection)urlRequest.openConnection(Proxy.NO_PROXY);
conn.getOutputStream();would also throw an ACE about SocketPermission when the applet tries to initiate Socket connection for the conn.getOutputStream() call.
Error Message(s):
java.security.AccessControlException: access denied (java.net.SocketPermission x.x.x.x:80 connect,resolve)
     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.checkConnect(Unknown Source)
     at java.net.Socket.connect(Unknown Source)
     at java.net.Socket.connect(Unknown Source)
     at java.net.Socket.<init>(Unknown Source)
     at java.net.Socket.<init>(Unknown Source)
     at HttpRequestor.start(HttpRequestor.java:16)
     at sun.applet.AppletPanel.run(Unknown Source)
     at java.lang.Thread.run(Unknown Source)
x.x.x.x being the resolved IP address of the origin server
Source code for an executable test case:
import java.applet.*;
import java.net.*;
public class HttpRequestor extends Applet
     public void start()
          try
               Socket sock = new Socket(Proxy.NO_PROXY);
               InetSocketAddress sockAddress = new InetSocketAddress(getCodeBase().getHost(), Integer.parseInt(getParameter("port")));
               sock.connect(sockAddress);
               String path = getParameter("path")+"&local="+sock.getLocalAddress().getHostAddress();
               String httpRequest = "GET "+path+" HTTP/1.0\r\nHost: "+getCodeBase().getHost()+"\r\n\r\n";
               sock.getOutputStream().write(httpRequest.getBytes());
               sock.getInputStream();
          catch (Exception e)
               e.printStackTrace();
Workaround:
The only possible workaround I could find is if the the applet can be loaded from IP-address in the codebase URL to avoid the unnecessary lookups by the SecurityManger.
This, however, maybe very difficult to use in most virtual-hosting environment, and IE browsers older than IE7 would most likely throw an error like "Class not found".
Example, if www.hostingsite.com resolves to 1.2.3.4
and http://1.2.3.4 goes to http://www.hostingsite.com
Change the codebase URL in the html embedding the applet, from:
"http://www.hostingsite.com/classes"
to:
"http://1.2.3.4/classes"
I also forgot to mention in my bug report that this would also be quite a big problem in server farms, where the origin-hostname resolves to multiple IP addressses and when those IP addresses do not all resolve back to the same hostname.

Different behavior of Apex (authentication schema) in 64 bits instance

Hello
First time poster after reading many, and not sure if this is Apex related or that I should create a SR on metalink on monday
We have Apex 3.0 installed in a development 10.2.0.1 instance on Win2003 Server (32 bits), the application uses an authentication schema which queries our ldap server (MS Active directory), which works fine.
Our production environment is Apex 3.0 in 10.2.0.1 (x64) on Win2003 (64 bits) 4GB of memory. The authentication (same as in development) takes a very long time (minutes). On the server we see (task manager) a huge increase of vm size usage when we press the login button. When idle it uses around 900MB of vm size, and increases with 5GB. This results in a non responsive db server for minutes. After authentication is done, vm size decreases to normal size.
In the enterprise manager (Active sessions waiting: Other) we see DBMS_LDAP: LDAP operation in the graph
steps taken
- Installed a fresh 10.2.0.1 instance (x64) on a Win2003 x64 vmware
- Installed the PL/SQL Web Toolkit 10.1.2.0.6
- Installed Apex 3.0
- Created authentication schema in the sample application based on ldap (tried DOMAIN_NAME\%LDAP_USER% and cn=%LDAP_USER%,ou=Users,dc=domain,dc=nl notation for the dn entry)
- log on to application authentication is slow, (ldap test tool shows same behavior)
- tried different domain controllers
- upgrade instance to 10.2.0.3
After the upgrade the behavior is as described above (slow, huge increase vm size, non responsive system).
Any help or ideas on this on.
Thank in advance
Roeland

Hello
Did some more testing this morning, on the 64bits test instance. Created a function with pl/sql developer (found on the forum)
create or replace function authenticate_aduser(
     p_username          in          varchar2,
     p_password          in          varchar2)
return boolean
is
     l_user               varchar2(256);
     l_ldap_server     varchar2(256)     := 'ldap_server';
     l_domain          varchar2(256)     := 'domain_name';
     l_ldap_port          number      := 389;
     l_retval          pls_integer;
     l_session          dbms_ldap.session;
     l_cnt               number;
begin
     l_user               := l_domain || '\' || p_username;
     l_session          := dbms_ldap.init( l_ldap_server, l_ldap_port ); -- start session
     l_retval          := dbms_ldap.simple_bind_s( l_session, l_user, p_password ); -- auth as user
     l_retval          := dbms_ldap.unbind_s( l_session ); -- unbind
     return true;
exception when others then
     l_retval := dbms_ldap.unbind_s( l_session );
     return false;
end;
- First time execution took 90 secondes, and showed the increase in vmsize in the task manager
- Second execution in 0,015 seconds
- Logged off
- Started a new session
- Again the first time execution took around 90 seconds and second execution was 0,017 seconds
Maybe I mis configured something or it's a bug. but it seems not Apex related, anyway going to create a SR
Regards Roeland

Mandatory fields & JHS different behavior for the very same situation

Hi,
There is an issue regarding mandatory fields in a form. Let me put it in this way:
Using this framework, obviously there are two ways for declaring a field as a mandatory one.
Firstly, at model level; checking the mandatory check box in the entity itself and secondly; at view level; one can set the required field in application definition.
The problem is; leaving a mandatory field empty; the user faces two different alert messages_.
Best Regards,
Yasaman Parandian
Edited by: Parandian on Oct 7, 2009 5:02 AM

Steven,
I can see a little bit of misunderstanding here; the point is that the javascript alert message is
the preferable behavior……And the question is if there is any way that I can change the JHS-level mandatory fields to pop up a javascript error message when left empty.
Thank you so much in advance,
Best Rgrds
Yasaman Parandian

Different behavior of installer depending on language

Related to my efforts to have a more informative installer message when someone re-runs the same installer (see this thread), I have encountered a language-specific difference that seems fishy.
If I build two installers of the same app, one with English as the installer language and the other with Simplified Chinese as the installer language, I get a different message if I run the English installer after the Chinese one. It refuses to install, and it says: Cannot Install. <appname> Files (higher version already installed)
That is a fairly informative message - more so than the usual "no software will be installed or removed" message that arises when one re-runs the exact same installer. But it does not seem correct.
The strange thing is that nominally the packages are the same version, and the English one was built after the Chinese one, so if there were any auto-incrementing that would not explain it. A peek at the setup.ini shows near-identical content with a different Package GUID in the [Distribution] section, a differernt [Localization} section and different GUIDs in the [Parts] section.
I'm planning to just ignore the behavior - but I'm a bit uneasy that it is a sign of potential bigger issues.
--Ian
Solved!
Go to Solution.

For some reason, when the versions of the installers are the same, but the installer languages are different, you get the same message that would be displayed if the installed application's version was higher. Note that, in general, if you try to install a different build of the distribution with the same version number, it will proceed with the installation. When the installer language is different, though, it does not install. I've filed a bug report (#323423) to address this inconsistency.
Mert A.
National Instruments

Different behavior in JWS vs. JDK environments

Similar Messages

Maybe you are looking for