OraTransplant

MetaLink

This page is mainly for my own reference to keep track of all TARs and bugs I’m interested in at MetaLink. Feel free to contact me if you have any questions regarding the TARs.

Technical Assistance Requests / Service Requests

6223193.993 - Time part of SelectInputDate increases by one hour in DST: We had an issue in which the time part of a SelectInputDate in ADF Faces increased by one hour each time you invoke the date picker (calendar) and select a date. This only happens for some dates during the year and seems related to daylight saving. A number of known bugs exist with similar behavior, but our case was not covered by any of these bugs. A new bug (5957009) was filed. This is marked fixed in JDev 10.1.3.3 and backports to 10.1.3.0 and 10.1.3.2 have been requested.
6180147.992 - Patch to workaround IE update to click active components: In February 2006 Oracle released a notice that Microsoft was going to update their Internet Explorer. Forms needs a JavaScript workaround to prevent the user needs to click on the Forms applet at startup to active the applet. Oracle promised updates for several Forms versions, but the patch number is not given in the document.
I asked Oracle support about it and they pointed me to MetaLink Note 357545.1 and Patch 5017052.
6134745.993 - How to use legacy_lifecycle with JInitiator 1.3.1.22: I’m working on a concept of using Oracle Forms as a . This concept uses the legacy_lifecycle parameter which instructs Sun JVM not to quit the applet when leaving the web page with the applet. In stead the applet is kept alive in the background and is restored when returning to a web page that includes the same applet. According to an Oracle document this behavior is also supported by JInitiator. Unfortunately I could not get it to work with JInitiator. So I opened a Service Request and found out that the JInitiator parameter for this behavior is jinit_appletcache. This parameter and the behavior was in JInitiator 1.1.8 (Forms 6i) but for some unclear reason was dropped in version 1.3.1 which is used for all subsequent Forms versions. So JInitiator 1.3.1 supports neither jinit_appletcache or legacy_lifecycle. To get this behavior you have to stick to Forms 6i with JInitiator 1.1.8 or use Sun JVM with the legacy_lifecycle parameter.
6066424.992 - Setting up webcache with TLS1.0 support: We had problems using some new browser to connect to applications and sites behind our Oracle WebCache version 10.1.2.0.2. As it turns out WebCache does not support TLS1.0 as a cypher suite for SSL encryption with a default installation. You can read all about the issue here. Oracle filed bug/enhancement request 5841589 to have a setting that allows WebCache to support SSL3.0 and TLS1.0 without also adding support for the less secure TLS2.0. The support engineer confirmed the fix will be in version 11 and we requested a backport to 10.1.2.0.2
6012055.993 - Comma gets stripped from decimal input in JSF page: We encountered an issue with JDeveloper version 10.1.3.0.0 where an input of “2,5″ would be interpreted as “25″ effectively stripping the comma. In the end it turns out that this occurs when we have a field with a format to include a decimal separator (e.g. ##.##), the language of the browser set to a European language that uses a comma as separator. In this case, the field is initially rendered with a comma. Submitting this value with the comma back to the application strips the comma from the input. I ended up writing a ServletFilter and RequestWrapper to strip the Accept-Language from the HTTP request. This causes the application to use English as the language in all technology layers. This fixed the problem and also made sure our international application always uses English regardless the browser settings. This is our desired behavior. The ServletFilter and RequestWrapper is described in a blog article and in Metalink Note 412158.1 as part of the Oracle Customer Knowledge Exchange
5971153.993 - What’s the status of bug 4522661: During our initial testing of using Sun JVM with Oracle Forms we encountered a new bug (4522661). This bug could cause Internet Explorer to hang when closing the browser window with the Forms applet. The problem only occurred using Sun JVM version 1.5, not with version 1.4. The problem ended up in the Forms 10.1.2.0.2 release notes and could not be fixed in the base release. In the end it turns out to be Sun JVM bug 5080386 and it should be fixed in Sun J2SE 6 which currently is not supported with Oracle Forms.
5953858.993 - ORA-600 [KZSPVCR:1] using proxy authentication: Using proxy authentication you can permit the proxy to connect as the specified user and to activate only the roles that are specified in a list. It turns out that if the list is longer than 16 roles, an ORA-600 can be raised when setting up the proxy connection. A new bug with ID 5660888 was filed for this issue. It is marked fixed in version 11.1 and we’ve requested a backport to database 10.2.0.2.0, which has been supplied
16597662.6 - How to use SSO with multiple URL path prefixes on the same hostname?: This SR was filed by a colleague of mine. We ran into problems since the success URL of mod_osse (the Apache module for Oracle single sign on) is hard coded at /osso_login_success. This results in problems since we have a load balancer (actually Webcache) that forwards requests based on the path in the URL. So http://www.eurotransplant.nl/app1 is forward to another origin server then http://www.eurotransplant.nl/app2. But both origin servers are protected by mod_osso and will forward the user to the central SSO sever. If that server approved the user it forwards the user to http://www.eurotransplant.nl/osso_login_success but there is no way for the load balancer to know to which origin server this request should be forwarded. This resulted in (hidden) note 390358.1 which describes a workaround if both origin servers run the same version. It also resulted in enhancement request 5512414 to make the osso-success URL configurable.
5267377.993 - Looping chain of synonyms hangs Forms: You can have a chaining loop of synonyms. The easiest way to create this is to have a public synonym pointing to an object with the same name as the public synonym. If that object does not exist, the synonym basically points to itself and you have a chaining loop of synonyms.
When you have a Form with a block that is based on this synonym, and you execute a query in Forms you’re Forms server process will loop with 100% CPU usage.
Bug 5140874 was filed and the SR was closed since it is a low priority issue
5267368.993 - How to set text color for disabled items: I asked support if it is possible to change the light gray color of disabled text items in Forms 10.1.2.0.2. As I already expected, there is no way to specify this color. There is only the colorScheme parameter to select a different color scheme. The foreground color of disabled items is one of the colors in such a scheme. You’ll have to pick a colorScheme with a foreground color for disabled items you like, or set the background color of such items so there is enough contrast with the disabled foreground color.
5254365.992 (open) - Single Sign Off not working: We have an issue with Single Sign Off not working. First visit a partner application that is protected by mos_osso. That redirects you to the SSO logon page. Once logged on, you are redirected to the partner application. No visit another partner application. You are not required to logon again, since you have a valid SSO session. Return to the first partner application and select the SSO Global Logoff link. That destroyes your SSO cookie and mod-osso cookie from that partner application server. Now return to the second partner application and it will grant you access although your SSO session has ended. It probably has something to do with the session ID from the second partner application still being valid. This is passed back and forth between the server and the client in a JSESSION cookie.
Somehow the second partner application isn’t notified of the logoff and hence that session is not destroyed.
16098382.6 - Backport request bug 5084833 to include in bundle for 10.2.0.2 win32: This is a request to include a fix for bug 5084833 (aka 5033218) to be included in the next patchbundle for 10.2.0.2 on Win32. The backport was approved and the fix should be included in 10.2.0.2 patch 2, which is the first non-CPU bundled patch for Windows.
5235422.993 (open) - Backport request 4908068 on HP-UX 10.2.0.2: This is a backport request to fix bug 4908068 in 10.2.0.2 on HP-UX. The backport has been approved and the fix is under development.
5235415.993 - Backport request 4908068 on win32: This is backport request to include a fix for bug 4908068 in the next patchbundle for 10.2.0.2 on Win32. The backport was approved, so the fix should be included in the next patchbundle for Windows
5238055.992 - How to cancel a running query: Since we upgraded from database 8.1.7 to 10.2 we also stopped using SQL*Plus 3.3 which was still installed on Citrix MetaFrame. Users started using SQL*Plus 10, but missed the option to cancel a long running query as they had in SQL*Plus v3.3. It turned out this is no longer possible since SQL*Plus v8.x since the version 8 OCI libraries are not multi threaded.
I guess they have to abandon SQL*Plus and switch over to Quest TOAD to get this.
5177911.993 - FRM-99999 Error 408 with deferred check constraint: When violating a deferred check constraint in Oracle Forms, you get a FRM-99999 Error 408 during commit. Also see a specific blog entry about commit time business rules in Oracle Forms. The handling of a violated deferred constraint by Oracle Forms ain’t pretty. The user first gets a message FRM-40400 x record(s) and applied. Just after this message Oracle Forms tries to commit to the database and runs into the violated constraint. This is shown to the user by the very cryptic “FRM-99999: Error 408 occured. See the release notes file (relnotes) for information about this error”
The support engineer came up with a great woraround. See the weblog entry for more details.
Bug 5091605 was created for the confusing messages that are raised by Oracle Forms. This bug was closed, but a new enhancement request 5101958 was filed to improve the FRM-40400 mesage.
5134355.992 (open) - Bug 3527043 seems not fixed in 10.2 ORA-6502: It looks like we are running into bug 3527043, where an ORA-06502: PL/SQL: numeric or value error: character string buffer too small is raised when inserting more than 1000 characters in a 4000 characters VARCHAR2 database item. This bug should be fixed in 10.2, but we still run into it after flushing the shared pool or waiting for a while.
You need some specific trigger code to reproduce this problem. When we just created or compiled the trigger the bug seems fixed. However, when you wait a while or flush the shared pool manually you still hit this bug (or something very simular).
Bug 5033218 was filed to track this problem. The bug is fixed for version 11 of the database. I’ve requested a backport to 10.2.0.2 for both HP-UX64 and Win32. The backport for HP-UX64 was approved and the backport request for Win32 has been filed as SR 16098382.6.
5084327.992 - Inline select for insert into not executed: A while ago I filed SR 5025221.992 when we had a problem that an INSERT on a table failed when using a column value of USER. If you have a trigger on this table that assigns a value to :new.column the initial value of :new.column in the trigger would be NULL. This turned out to be known bug 4547458 and a one-off patch was available.
Now we’re running into a very similar problem. First you have to have a trigger on a table that assigns a value to :new.column. In our case the Designer TAPI generated triggers do this (assign :new.column to a variable, make some changes to the variable and then assign the variable back to :new.column). Then you have to do an INSERT INTO TABLE VALUES where you use a SELECT as one of the values, eg insert into table t values ((select 1 from dual));. This will leave the value of :new.column NULL in the trigger. Since our trigger assigns this to a variable and then assigns the variable back to :new.column, we end up with NULL being inserted in the database.
A new bug 5001249 has been filed for this issue. Let’s hope they can fix this as soon as possible. This is a showstopper for our 8.1.7 to 10.2 migration. Patches for this issue are now available. See the patches/workarounds section on this page
5033220.992 - Creating custom change password page: Note 312059.1 describes why a custom change password page on SSO won’t work when the policy ‘Reset password on Login’ is enabled. Basically when a password is expired and the user changes it, the password change is done by an admin user. This triggers the ‘Reset password on Login’ policy and the user is required to change the password again.
To circumvent this problem I’m trying to write my own ChangePassword servlet. To not trigger the ‘Reset password on Login’ policy I’m logging on to the OID/LDAP server with the username and old password supplied on the change password page. This works ok in most circumstances. The only time it doesn’t work is when the password was changed by an admin. At that time, the user is not allowed to read any information from OID/LDAP which is required to change the password: LDAP: error code 53 - Password Policy Error :9009: GSL_PWDMUSTCHANGE_EXCP :Your Password has been reset; You must change your password before performing other operations. So I would have to go back to an admin user to make the change but that would give me the same problems as in Note 312059.1.
I’ve asked Oracle support to come up with some simple Java code to change a reset password without tripping the ‘Reset password on Login’ policy.
5032612.992 - Backport request for bug 4547458: In SR 5025221.992 we asked for a backport of bugfix 4547458 to HP-UX. This SR (5032612.992) was opened to ask for a backport fix to Windows since that is our primary development plaform
5026104.993 - Change password page error codes: The Application Server documentation on using deployment specific pages for SSO has a list of errors you can get when changing a password. The list is included in the v9.0.4 and v10.1.2 documentation.
It appears that the list is not correct. Some errors are misspelled and I get some additional errors when changing a password that are not on the list. As it turns out a documentation bug 4738777 was already filed. I’ve asked support for a full list of the correct error codes. Let’s hope they will publish that as a metalink note.
5025221.992 - Default column of table not filled: In database 10.2.0.1.0 we ran into a problem with a database trigger. When a column has a default value of USER, the :new.column variable is NULL when the trigger contains an assignment to that same :new.column. Something like:
v_a := :new.a; dbms_output.put_line(’new.a at start trigger ‘||:new.a); :new.a:=v_a;
When you remove the last assignment: :new.a:=v_a;, all runs well. A new bug 4932337 has been filed for this. That bug has been closed as being a duplicate of 4547458, which is fixed in the v11.x codebase of the database. I’ve asked for a backport fix to 10.2 on HP-UX and Windows.
In the end, a separate SR (5032612.992) was filed for the backport request to Windows
15899900.6 - RDA hangs: I was asked by Oracle Support to run RDA to collect information about my 10.2 database on Windows. However, it hangs during the PERF module. Oracle created this SR to handle the issue.
4993480.993 - ORA-6502 when trigger references : old value: I ran into an issue with Oracle database v10.2.0.1.0. On HP-UX an update fails with ORA-06502: PL/SQL: numeric or value error: number precision too large in a trigger that just gets a : old.column value. When running the same statement on Windows the whole session crashes with ORA-07445: exception encountered: core dump [ACCESS_VIOLATION] [__VInfreq___PGOSF374__upduaw+2158] [PC:0x2E89BB6] [ADDR:0x0] [UNABLE_TO_READ] []
The user performs an update on a view. The underlying table of that view has the trigger that fails. The view itself references another view in its WHERE clause. The underlying table of the references view has a primary key constraint set on the column that is referenced in the WHERE clause of the first view.
This is not a very uncommon situation and I’m wondering why this crashes Oracle. Let’s hope Oracle support can come up with an answer quickly. Oracle closed this SR as a duplicate of the spin off 15863162.6. Work on that other SR continues.
15863162.6 - ORA-7445 [access_violation] [__vinfreq___pgosf374__upduaw+2158]: This is a spin off service request 4993480.993 for the Windows platform. Oracle decided it is better to process it as two different service requests.
Oracle decided to close the original SR 4993480.993 and change this SR from Windows to HP-UX. I guess they’re doing there best to confuse things. Bug 4904890 was filed for HP-UX.
4975938.993 - Enh.req.: keep frmcmp silent when running in batch mode: When you call the Forms compiler on Windows (frmcmp) with batch=yes and window_state=minimized it still starts an application in the taskbar. This application takes focus meaning your current application looses focus. This can be quite annoying when you’re running an automated build which builds a couple of hundred forms. A new forms compiler keeps popping up that takes focus. This means you cannot continue doing any other work while the build is running. Nothing major, but I thought I take my chances and file an enhancement request. This is filed as “bug” 4892033
4867490.993 - Export of char_semantics 10.2 database fails with ora-932 exp-56: The traditional (non DataPump) export tool failed exporting a database with NLS_LENGTH_SEMANTICS set to CHAR. As it turns out this is caused by setting the NLS_LENGTH_SEMANTICS to CHAR before creating the database. You should only do this after creating the database. I’ve posted a specific blog entry for this. It wil also be included in a number of Notes on MetaLink, mainly 144808.1. A new bug 4886376 was filed to have this fixed.
4866151.992 - impdp grants roles to user creating them exceeding max enabled roles: I tried a full export of our 10.2 database using expdp. I then imported this into a fresh 10.2 database. That resulted in ORA-31625: Schema SYSTEM is needed to import this object, but is unaccessible and ORA-28031: maximum of 148 enabled roles exceeded. Our source database has more then 148 roles defined. They are all created using the SYSTEM schema with the IMPDP. Oracle by default grants a role being created to the user creating it. This resulted in more than 148 roles being granted to SYSTEM. This TAR resulted in the creation of bug 4744469.
4788473.992 - How to create and call custom netca.rsp: With build automation using Ant I would like a developer to setup only a minimal workstation. The rest should be handled by Ant. This also means a developer would only install the Oracle database software without a starter database. The database will be created by Ant. Since the Oracle Installer doesn’t create a database, it also does not run the Network Configuration Assistant to create a default network config. So I asked support how to automate this. The answer was netca /silent /responseFile "c:\response\netca.rsp" /log "c:\response\netca.log" This will create the default confif files (sqlnet.ora, tnsnames.ora and listener.ora) and will create the OracleXXXTNSListener service on Windows. Look for the example .rsp file in your ORACLE_HOME for some documentation.
4667246.992 - How to call custom change password page: We created deployement specific logon, logoff and change password pages for our Single Sign On environment. We installed them as documented in the SSO documentation. The change password page is called when a password expires, but I would also like to offer a user the possibility to change her password at will. How do I need to call this custom change password page?
The change password page needs a site2pstoretoken that has to be generated by Oracle first. Note 290445.1 explains how to do this using Portal. We couldn’t use this solution since we do not use Portal. It turned out that it is just not possible. We will have to create our own Change Password page and have it interact directly with Oracle Internet Directory using LDAP. That’s a shame
4657475.993 - Menu being shown is not the actual menu: Sometimes our application would display a wrong menu. It turned out the cause was having windows with the name ROOT_WINDOW. The got a special treatment before the MDI (MultiDocument Interface) and that caused some strange behaviour after migrating from Client-Server to the web.
4651578.992 - forms90/java/ewt/scrolling/mousewheelhandler.class not found: When installing Forms patchset 9.0.4.2.0 we found out that MouseWheelHandler.class was missing from the f90all.jar. The client tries to retrieve this file sometimes more than a 100 times. This delays the application considerably when running over SSL, because of bug 4259148. That bug causes an exception to be thrown in the Java console when downloading a non-existant file over SSL. On low end hardware all this exception throwing takes quite some time.
In the end Oracel released patch 4259148 that fixed the issue for version 9.0.4.2.0. It’s basically the same as extracting the f90all.jar yourself and adding the missing file from a frmall.jar from version 10.1.2.0.2. I’ve written a seperate blog entry about this issue.
4648156.992 - Function key not allowed when opening form from menu FRM-41009: Sometimes we would get stuck in our Forms application. Whatever you did resulted in a FRM-41009 Function key not allowed error. We managed to create a small reproducable testcase. Bug 4598189 was filed. Further investigation revealed the problem is caused by having a window with the name ROOT_WINDOW. We also had another issue with windows with this name, see TAR 4657475.993.
It turns out that in the old days (before the multi-document interface in Windows), a window with the name ROOT_WINDOW got some special treatment. This is no longer necessary in a MDI interface, but as it turns our it can still lead to problems. The simplest solution is just to not use windows named ROOT_WINDOW. We still had a lot of them from the SQL*Forms 3.0 era.
The TAR also resulted in the creation of Note 334210.1 explaining the issue and resolution.
4610541.992 - Illegal restricted procedure Synchronize in when-validate-item: We had a reproducable testcase where a Synchronize call in a When-Validate-Item trigger would raise a FRM-40737 Illegal restricted procedure SYNCHRONIZE in when-validate-item. Synchronize is not a restricted built-in so this error is plain wrong. I could find several bugs relating to Synchronize being treated as a restricted built-in, but they are all supposed to be fixed back in the v6.x era. The first suggestion by the support engineer was to remove the calls to Synchronize whenever possible as they have an impact on performance. In the old days (v4.x and v5.x) you would need to call Synchronize every now and then to workaround bugs. Since most of them should be solved in the latest products, you can try to just remove the Synchronize calls.
The service request did result in filing a new bug: 4576837. That bug was closed as being a duplicate of bug 2509290. This latest bug was originally filed against v6.0.8.19.2 but it is still open. It had been suspended for a while because not enough information was available. With our new testcase it has been reopenend and is currently being fixed.
4543902.992 - Null d2fob pointer when converting XML to FMB: After installing Forms patch v9.0.4.2.0 on top of v9.0.4.1.0 we could no longer convert a number of Forms to XML files and back. The problem lies in converting the XML files back to FMB. Initially bug 4504586 was filed, but it was closed as not being a bug. We investigated the issue a bit more and a new bug (4527773) was filed. The problem lies in having a subclassed LOV which has an overridden ColumnMapping property. The bug has been fixed in the coming v11.x release of Oracle Forms. A backport request for v10.1.2 has been filed.
4522876.992 - Performance degredation when large amount of data on clipboard: On every item navigation Forms checks the type of data on the Windows clipboard to determine if the Edit > Paste command should be enabled. Due to a Sun Java bug this checking of the data type retrieves the entire content of the clipboard. When having lots of data on the clipboard (e.g. a screenshot) this can cause high memory consumption and considerable delays on low end hardware. This was fixed in v6.0.8.25 as bug 3130709, but somehow didn’t make it in the v9.x branch. The bug has been re-filed as 4520121. The issue is planned to be fixed in v10.1.2.1. A backport fix for v9.0.4.2.0 has been released as patch 4599915. A backport fix for v10.1.2.0.2 has been requested as bug 4684569.This has been released as patch 4520121.
I’ve got a separate blog entry explaining this issue and the fixes.
4446709.994 - Cursor sticks to first item on mouse navigation: Users complained about sometimes getting stuck in our application. You can no longer navigate to other items using the mouse. You can still navigate using the keyboard (Tab). In the end, we found a reproducable testcase that was submitted to Oracle. The problem was caused by a requery in during the commit processing. That code would see what the current block is. It would then navigate to the master block, execute the query and go back to the original block. During the requery a trigger disabled a tabsheet that was subsequently re-enabled by another trigger during the same query processing. If the user was in the block that is on the (shortly) disabled tabsheet, he would run into the problem. But only when using Sun JPI v1.5. Version 1.4 did not have the issue. Apparently the disabling and re-enabling of the tab confused the applet running on the client. The workaround is to include a Synchronize just before the final go_block statement. That fixed the issue. Bug 4683804 was filed to come up with a final solution.
We also discovered that we had other simular cases where the user was using Sun JPI v1.4 and we are not disabling items. This turned out to be another known bug: 3973176

Bugs

5512414- ENH: mod_osso support for application-specific success_url: Created as a result of SR 16597662.6
5841589- Provide setting in webcache to restrict SSL protocols to SSLv3 and TLS1.0: Created as a result of SR 6066424.992.
5957009 - Time part of SelectInputDate increases by 1 hours in DST: Created as a result of SR 6223193.993. The issue is marked fixed for JDeveloper/ADF version 10.1.3.3 and backports to 10.1.3.0 and 10.1.3.2 have been requested
5660888 - ORA-600 [KZSPVCR:1] using proxy authentication: Created as a result of SR 5953858.993. The issue is marked fixed for version 11.1 and we’ve requested a backport to 10.2.0.2, which has been supplied
5140874 - Querying against a circular ref. synonym hangs Forms: Created as a result of SR 5267377.993.
5101958 - Misleading ‘FRM-40400 1 record applied and saved’ message: Enhancement request to improve FRM-40400 message. See SR/TAR 5177911.993
5091605- FRM-40400:Transaction complete: x records saved and applied’ message although a deferred constraint is violated: See SR/TAR 5177911.993
5033218 - Intermittent ORA-6502 if trigger modifies :new value with length 1001 on insert: See SR/TAR 5134355.992
5001249 - Inline select for insert into not executed: This bug was filed as a result of TAR 5084327.992. See that TAR for a more detailed description of the issue. The bug is basically the same as bug 4908068 which is filed for Windows. This bug is fixed in version 11 of the database. I’ll request a backport fix to 10.2 on HP-UX and Windows.
4908068 - Subquery doesn’t find value that is available: See TAR 5084327.992 and bug 5001249. This is the basebug that was already filed at Oracle and is marked fixed in version 11 of the database.
4932337 - Default column of table not filled after modified in trigger: This bug was filed as a result of TAR 5025221.992. It has been closed as a duplicate of bug 4547458.
4904890 - ORA-7445 [access_violation] [__vinfreq___pgosf374__upduaw+2158]: See TAR 15863162.6
4892033 - Enhancement request: Not available on Metalink (yet). To be described
4886376 - invalid ku$_xmlschema_view creating db with nls_length_semantics=char: To be described
4744469 - impdp grants all roles to user creating them exceeding max enabled roles: To be described (not a bug)
4738777 - Doc: missing/wrong error codes for custom change password page 10.1.2: The list of error codes for the custom change password page in the v10.1.2 documentation is wrong and incomplete. This is a documentation bug I ran into. Also see service request 5026104.993
4683804 - Sticky cursor with JPI 1.5 but not with JPI 1.4 not JInitiator: To be described
4598189 - FRM-41009 “function key not allowed” when opening a form(no menu): To be described (not a bug)
4576837 - Illegal restricted procedure Synchronize in when-validate-item FRM-40737: To be described
4547458 - Insert using ‘USER’ SQL function fails with ora-1400 when using before trigger: Service request 5025221.992 resulted in a new bug being filed under number 4904890. That bug was closed as being a duplicate of 4547458
4527773 - Null d2fob pointer converting xml to fmb-Column mapping overridden: To be described
4522661 - JDK1.5:Window XP:IE:Browser hangs on exiting form & closing browser: To be described (not a bug)
4520121 - Performance degredation when large amount of data on clipboard: Bug that was filed to get the fix for bug 3130709 in the v9.x/v10.x branch
4504586 - Error - an exception has been encountered null d2fob pointer: To be described
4259148 - Running form with JPI over SSL throws exception error in java console: Unpublished bug about missing MouseWheelHandler.class in f90all.jar. Also see TAR 4651578.992. Oracle did release a patch for this unpublished bug. This patch has the same number 4259148
3973176 - Sticky cursor with Forms deployed with Sun JRE: To be described
3208897 - call_form(do_replace) leaves menu of called form if the form is exited: To be described
3130709 - Performance degrades with clip board having large amount of data: Bug that was fixed in v6.0.8.25 but the fix never made it into the v9.x branch
2509290 - Recieve FRM-40737 error while scrolling records in webforms: To be described

Notes

339938.1 - Full Export 10.2.0.1 Aborts With EXP-56 ORA-932 (Inconsistent Datatypes): To be described
334210.1 - Frm-41009 Function Key Not Allowed When Opening Form From Menu: To be described
312059.1 - Reset passwd on login feature not working with custom change pword page: To be described
290445.1 - Changes to the Login Portlet implementation for Portal 9.0.4.1/10.1.2: To be described
252273.1 - What is ‘STARTUP MIGRATE’?: To be described
144808.1 - Examples and limits of BYTE and CHAR semantics usage: To be described
1033387.6 - Menues does not Displayed After Return from Calling Form: Note that describes the problem we logged as TAR 4657475.993. This was fixed in Forms 4.5.7.6.1 and 5.0.4.16.0. We encountered it again in Forms 9.0.4.2.0 on the web. The workaround was enough for us.

Patches / Workarounds

Bug	SR	Product	Description	Win32	HP-UX64
5660888	5953858.993	Database 10.2.0.2.0	ORA-600 proxy authentication with 17 roles	- Fixed in v11	- 5660888 on 10.2.0.2.0 - Fixed in v11
5033218	5134355.992	Database 10.2.0.x	ORA-6502 if :new.value with length 1001	- Keep triggers in shared pool - 10.2.0.2.1 patchset - Planned for 10.2.0.3 - Fixed in v11	- Keep triggers in shared pool - 5033218 on 10.2.0.2.0 - Fixed in v11
4908068	5084327.992	Database 10.2.0.x	Subquery not executed on insert	- 10.2.0.1.5 patchset - 10.2.0.2.1 patchset - Fixed in v11	- 4908068 on 10.2.0.1.0 - 5064500 on 10.2.0.1.0 (merge with 4547458) - 4908068 on 10.2.0.2.0 - Fixed in v11
4904890	4993480.993 15863162.6	Database 10.2.0.x	ora-6502 or crash referencing :new.value in trigger	“_optimizer_push_pred_cost_based” = FALSE - Fixed in v11	“_optimizer_push_pred_cost_based” = FALSE - 4904890 on 10.2.0.1.0 - 4904890 on 10.2.0.2.0 - Fixed in v11
4547458	5025221.992	Database 10.2.0.1.0	Insert using SQL function USER fails	- BP 10.2.0.1 denied - 10.2.0.2.0 patchset	- 4547458 on 10.2.0.1.0 - 5064500 on 10.2.0.1.0 (merge with 4908068) - 10.2.0.2.0 patchset
4527773	4543902.992	Forms 9.0.4.2 & 10.1.2.0.2	Null d2fob pointer when converting XML to FMB	- 4527773 on 10.1.2.0.2 - Fixed in v11	- 4527773 on 10.1.2.0.2 - Fixed in v11
4520121	4522876.992	Forms 9.0.4.x & 10.1.2.0.2	Performance degrades when large amount of data on clipboard	- 4599915 on 9.0.4.2.0 - 4520121 on 10.1.2.0.2 - Fixed in v11	- 4599915 on 9.0.4.2.0 - 4520121 on 10.1.2.0.2 - Fixed in v11

28 Comments

TrackBack URI

Hi Wilfred

Skimming the entries in this post, I stumbled on “4446709.994 - Cursor sticks to first item on mouse navigation”. This sounds similar, if not VERY similar, to some problems that we are currently experiencing with our clients, who we have currently forced to run on JRE 1.4.2 because of bug #5668071, that I have initiated. I have had this bug escalated, and I hope to upgrade our customers to JRE 1.6, as soon as this bug is fixed.

In brief, this bug covers an issue, where the focus on the form dies after doing a WEBUTIL_FILE.AS_TO_CLIENT_WITH_PROGRESS on a larger file, 967kB is what I have tested it with. With a small file on, say 1kb, this does not appear. The funny about this is: the only Java version, where this error does NOT happen, is 1.4.2 - it happens with JInitiator (1.3), 1.5 and 1.6…

I don’t know, if you can access this bug 5668071 on MetaLink, but if you can - does this sound familiar to you? Also, I would like to ask you - are the issues with 4446709.994 solved in JRE 1.6? Have you had a chance to test this?

Comment by Jacob — 11 January 2007 #
Also found this
4683804 - Sticky cursor with JPI 1.5 but not with JPI 1.4 not JInitiator

Let me know, if you get a chance to test this with 1.6. It also sounds quite familiar to me.

Comment by Jacob — 11 January 2007 #
The Sticky Cursor problem has been bugging is ever since we migrated to the web. In September 2005 we managed to get a reproducible testcase and I filed service request 4446709.994. This only reproduced using JVM 1.5. It would not reproduce on JVM 1.4. Support didn’t really want to handle the SR, since JVM 1.5 wasn’t officially supported yet. Fortunately Grant Ronald was so friendly to immediately update the Client Statement of Direction on OTN and added support for JVM 1.5

This allowed Oracle support to continue working on the issue. Bug 4683804 was filed on our behalf. We found out you run into the problem if during navigational triggers you disable a tab page, then enable the same tab page again and then navigate to an item on the tabpage. Somehow the disabling and re-enabling of the tab caused the problem. Issuing a SYNCHRONIZE after enabling the tabpage was a successful workaround. This workaround sufficed for us since it only affected a single form. I never understood why development closed the bug because of missing information. They had a 100% reproducible testcase.

In the description of bug 4683804, another bug was mentioned with ID 3973176. The title of that bug seemed to indicate a similar sticky cursor bug. This time it is triggered by switching back and forth between Forms and another application. We also encountered this. It seems to be fixed in Sun JVM 1.4.2_13 and 1.5.0_10.

Internally we upgraded to 1.4.2_13 and the problem did disappear. We do not use 1.5.0 because of other problems (hanging browser when closing Internet Explorer).

So, it seems like there are a whole lot of ways you can loose mouse navigation in a form. If you search metalink you can also find others. Your bug seems related to the use of webutil, which we don’t use in our applications. You can take a look at bug 5754896 on Metalink. That testcase also includes switching to another application while Forms is busy. Perhaps the file transfer is implemented in such a way that it looks like switching to another application from the Forms standpoint. Give it a try with Sun JVM 1.4.2_13 and perhaps also throw in the suggested one-off patch 5677148 from the bug description. That’s a combined patch for bugs 5624261 and 5474223 (both unpublished).

Comment by wilfred — 14 January 2007 #
Hello Wilfred

I thought, that I would let you know, that I am currently testing an unofficial beta of JRE 1.5.0_11, which seems to fix all of the issues with 1.5, that I encountered, except one (an issue with java.awt.PrinterJob.printDialog()), that I have reported today to Oracle, and that I hope, that Oracle will pass on to Sun to have them fix the problem in the final version of 1.5 Update 11. It also takes care of the bug, that I posted about earlier, as long as we talk 1.5.

Unfortunately, this bug also occurs on 1.6, but I have reached the point myself, where all I want is a JRE, that simply *WORKS*, regardless of being 1.5 or 1.6 - and if I can get fully working 1.5 release, that will be enough for me for now. I have asked Oracle to log a bug with Sun to make them fix the issue in a forthcoming 1.6 patch also.

This beta is available on MetaLink as patch #5512094, but you need a password to download it. I don’t think, that I should post that password here, but create an SR yourself and ask for it, and you will get it, I am sure.

My problem thereby somehow seems to be related to bug #5512094: NO SYNCHRONOUS FOCUS TRANSFER DURING SWITCHING WINDOWS. Somehow I am sure, that you have encountered this bug before yourself - and you should be aware, that similar problems are also present still in 1.6…

I hope, that I can get you to take a look at this JRE beta sometime, if you get a chance. Would be great to have some additional feedback to provide to Oracle.

Thanks in advance.

Jacob

Comment by Jacob — 25 January 2007 #
Hi Wilfred…you know that I’ve run my Forms application with JRE instead of JInitiator. My Forms version is 9.0.4.3 and JRE version is 1.4.2_13. In fact, I was recommended to use this version of JRE but now to have more previleges out of JRE I’d like to use JRE 5.0. Do you think there might cause any problem using JRE 5.0 instead of JRE 1.4.2 with Froms version 9.0.4.3?

Comment by Rashedul Hasan — 29 January 2007 #
The main reason we advice our users not to use JRE 5.0 is the issue mentioned in section 4.1.4.4 of the Forms 10gR2 release notes. But from the comments above you can see that Jacob has a lot of other issues with JRE 5.0. In the end it’s your own choice which JRE you want to use and support, but remember that Oracle does not support JRE 5.0 with Forms 10gR1 (9.0.4), so you’re on your own.
Since you’re going to run without official support anyhow, you might even consider J2SE 6. There is a problem with some missing classes, but there seems to be a workaround.

Comment by wilfred — 29 January 2007 #
I must agree with Wilfred and say, that if you want a full supported version, 1.4.2_13 is the way to go right now, no matter if you run 10gR1 or 10gR2 - and since you are still running 10gR1 anyway, going 1.5 would be a very bad choice in my opinion, both because of the known issues, and for the lack of official support.

This version also appears to be the best working version out there right now. But due to application requirements, I’m forced to move to 1.5 or higher soon, and since Sun doesn’t seem to be able to get 1.5 right, I decided to go and play around with 1.6, even thought it’s unsupported. The beta I mentioned earlier is a step in the right direction, but having found yet another issue since my last comment, my hopes are not that big for 1.5.0_11 - but it will be worth a test, when it’s final.

Or - you might want to take a chance and begin playing around with JRE 1.6, like we did. But if you do, I would strongly recommend, that you move to 10.1.2.0.2 (10gR2) of Forms. I must advise though - apart from the mentioned issue with the missing classes - JRE 1.6 still has window focusing issues, unfortunately. It seems like bug #5512094 has survived into 1.6. But you may or may not be affected by this, depending on how your application is built.

Comment by Jacob — 29 January 2007 #
Hi Wilfred

Thanks…I’ve successfully downloaded and installed JPI on client PC from server like JInitiator using jpi_codebase as you suggested. In fact, I had to run my application using javaplugin because of a vital issue required for my application. I tell you that now…I’m developing a TextEditor where BMP images will be embedded and for embedding I had to use Java Advanced Imaging (JAI) tool. So, when I came to use my program on Oracle Forms, I found that image embedding isn’t possible here because JAI tool is compatible with JRE 1.4. This is why, I used JRE to run Forms Application. And with your help I’ve also been able to download and install JPI on client PC. Well….If I can now download the JAI tool on client PC after JRE gets downloaded and installed, it would be great. As the JRE needs to be upgraded with JAI tool for advanced Imaging operation, I’d like to provide this facility to client.

I’ve already tried it different ways like putting the JAI exe after JRE exe in jpi_codebase with coma and semicolon etc. Now if you could show me a way to accomplish this, I’d be very much benefitted.

Regards

Rashed

Comment by Rashedul Hasan — 2 February 2007 #
I have no experience with JAI at all, let alone auto installing it. But I did find a reference on auto installation in the documentation. Perhaps that can help you.

Comment by wilfred — 2 February 2007 #
Hi Wilfred….I’m facing a strange problem with JInitiator. I configured my server and ran my application with both JInitiator and JPI (using webutiljini.htm and webutiljpi.htm). When I run with JInitiator, a “Jninitiator Security Warning” messagebox appears saying “Unable to identify certificate - code will be treated as unsigned”. I need to let you know that I configured another server before this one and there the configuration is same but no such warning is shown switching between webutiljini.htm and webutiljpi.htm. I haven’t found any configuration mismatch yet in these two servers. If you do have any idea on this, please let me know.

Regards

Rashed

Comment by Rashedul Hasan — 5 February 2007 #
Which JAR files are included in your ARCHIVE parameter for the applet? It appears that at least one of them is signed with a certificate that is not recognized by JInitiator.

Be aware that JInitiator only has a very limited number of root certificates by default. So, if you signed a custom JAR with your custom certificate you got from Verisign or any other provider, you run the risk that the Verisign (or other) root certificate that was used for your certificate is not known by JInitiator. See this blog for more information.

This all would not explain why it works on the other server, but perhaps there is a difference in the used JAR files?

Comment by wilfred — 5 February 2007 #
Hi Wilfred…I’m facing some problems running my Forms application using JRE/JPI. My Forms version is 10g release 1 and JRE version is 1.4.2_13. I’ve upgraded my Application Server with patch to make it run smoothly with JRE. Patch ID is p4948949_9043_WINNT (patch 3). So my upgraded Forms version is 9.0.4.3.

Problems are as follows:

** There’s a problem in Font, it’s not appearing on application as configured. So I want to unistall the patch and restore my Forms version to previous version but I don’t know how to uninstall this patch.

** Another issue which seems to be JRE related problem is I can’t use mouse to select control on my Forms running application. I’ve to press TAB key to navigate through controls such as TEXT_ITEM. When I’m running application using JInitiator, this problem doesn’t occur, I can click on any control of Forms , so I think it’s JRE related problem.

Ours is a large scale application…this is why I’m in great problem now. Could someone help me resolve these issues?

Regards

Rashed

Comment by Rashedul Hasan — 6 February 2007 #
You might have to take a look at the font mappings in ORACLE_HOME/forms/java/oracle/forms/registry/registry.dat In that file you setup the mappings between the font names used in the Form and the font names to use in the Java VM to display. Most of the times your Forms originate from the Windows platform and use Windows font names. You have to map these to Java font names to enable the virtual machine to display them.

I’m not sure if it is possible to deinstall patchsets. You could try to start the Oracle Universal Installer and see if you can do it from there. But your probably best of to first ask advice from Oracle Support at http://metalink.oracle.com

The problem you describe with the mouse navigation is what is known as the sticky cursor problem. We had several different causes for the same problem. Just search for sticky cursor on my blog to find a few. You can also search for sticky cursor on metalink and you’ll find a whole number of documents.

We mostly have these problems with higher versions of Sun JRE and they don’t reproduce 100% of the time. For example when switching between Forms and other applications this might sometimes happen but not always. The only case we had that could reproduce 100% of the time was when we disabled a tab page in a trigger and re-enabled it shortly after that in another trigger that fired at the same time. The code then tried to navigate to an item on this disabled and re-enabled tab. That caused the loss of mouse navigation. Adding a SYNCHRONIZE after the re-enabling of the tab fixed the problem for us.

So, for the sticky cursor see if you can find a solution on the net or metalink. Also check if it happens with a newly crafted form or if it specific to your forms. If the latter is the case keep stripping program units and triggers from the form until the problem disappears. This can give you a hint on what’s going on and how you can workaround it.

Comment by wilfred — 6 February 2007 #
Hi Wilfred….maybe you answered one of my postings in Oracle Forms blog regarding Opatch. I have posted my problem in detail there but you didn’t answer yet. Would you please visit there and provide me the workaround? The link is as follows:
http://forums.oracle.com/forums/thread.jspa?messageID=1677895&#1677895

Regards

Rashed

Comment by Rashedul Hasan — 9 February 2007 #
Rashedul,

Yes that was me, but I attended a seminar yesterday and did not have access to my computer. I just answered on the OTN Forum.

Comment by wilfred — 9 February 2007 #
Hi Wilfred…I can make a Javabean as a button creating a JButton in a PJC class and using dispatchCustomEvent()….and on Forms I just write little code in When-Custom-Item_Event trigger to make the button functionality ok such as button press event. Now I want to do the same for a JApplet or JFrame. Is it possible to make the JavaBean a Java Applet or Frame? If so…how? Could you tell me about this?

Comment by Rashedul Hasan — 19 February 2007 #
Hi Wilfred

Running my Oracle Forms application wiith JavaPlugin (JRE 1.4.2_13) I encountered a strange problem…thats is when I close my application, the JRE icon doesn’t get vanished from Task Bar. Could you help me solve this problem?

Regards

Rashed

Comment by Rashedul Hasan — 24 February 2007 #
Rashed,

That is expected behavior. The Java Virtual Machine that is started as a child of your browser doesn’t close until the last browser window of that same session is closed.

Comment by wilfred — 24 February 2007 #
Please, somebody, could you teach me step by step how create a database with tablespaces, datafiles, schemas and users in oracle 8i? I have not oem, I have to do it using commands.thanks

Comment by Christian — 21 June 2007 #
Christian,

Setting up a new database with content (tables, etc.) is a whole lot of work. I can’t give you a simple list of steps to do. You’re probably best of by start reading the documentation at http://docs.oracle.com

Also, are you sure you want to go with the outdated version 8i if you’re setting up a new database. I would advice you to go with the latest version if you’re doing a new install. In fact, it’s much easier with 10gR2. Just download Oracle Database XE from OTN at http://otn.oracle.com. It comes pre-configured and has an extra easy install. And it’s totally free!

Comment by wilfred — 21 June 2007 #
Hi,

We are using Oracle 10g application server for our application. When we press tab to navigate to another control, the application starts hanging. Has anyone faced this problem or is there any workaround for this problem

Comment by Viji — 10 June 2008 #
Hi Viji,

Sounds like one of the focus issues with Oracle Forms. There are several reasons why Forms messes up the focus handling and you can no longer navigate correctly to other items. You’re best of contacting Oracle Support at http://metalink.oracle.com to have your problem analyzed.

Comment by Wilfred van der Deijl — 15 June 2008 #
ı dont understand

Comment by tanjant2x — 23 October 2008 #
Urgent assistance needed.

Symptoms: AR system getting FRM-40735 and FRM-47023 when loading forms through application servers. When accessing forms directly on DB server there are not problems. Any advice?

Comment by Tommy — 11 February 2009 #
Hi,
Can we change jre version of Forms 6i which uses 1.6 and i require 1.4 to use .
Thnaks in advance.

Comment by sumit — 8 June 2009 #
card-counting-blackjack.com. Blackjack card counting is the way to beat the casino. You count the cards played, and keep track of when the odds of the game shift based on the remaining cards. You can actually change the advantage to yourself instead of the casino. The free online game teaches you how to count cards.

Comment by blackjack — 15 June 2009 #
Pozdrowionka z Polski
Zapraszam na stronę >> Bieszczady >> http://www.it.bieszczadzkie.pl

Comment by Bieszczady — 28 July 2009 #
thank you number one medyum turkey

Comment by medyum — 2 November 2009 #

Trackback/Pingbacks:

hostname missing or wrong error

[...] codes for the custom change password page in the v10.1.2 documentation is wrong and incomplete.http://www.oratransplant.nl/metalink/Domain GuessingDomain Guessing intercepts the DNS "hostname not found" error, and resends [...]
Designer Checks

Designer Checks... Thanks for creating a really informative site. It's more than most people do! The OraTransplant " MetaLink was very helpful indeed. Jeremy....

Sorry, the comment form is closed at this time.