3.1EA3 Worksheet parser bug

Hello, I found this bug yesterday while writing this thread:
3.1EA3 Members of package body not listed in the tree - Still not fixed!!!
Simply put this into your worksheet
CREATE LIBRARY TEST_DLL AS '/a/b/c/x.dll';
SELECT * FROM USER_LIBRARIES;
DROP LIBRARY TEST_DLL;Part of first line is underlined highlighting there is some error in this statement even there isn't.
Place the cursor on the first line and pres Ctrl + Enter. The parser is probably confused by those slashes and "executes" whole worksheet... Now highlight second line and press Ctrl + Enter and look what was created.

Hello Vadim,
I think you are missing the point. The point is that you make a mistake in declaring new package body or spec and when trying to compile it, you don't get any warning or error. The procedure how you create new spec is irelevant...
So once again.
1) Create new empty file with name TEST_PACKAGE.pks
2) Open it in SQL Developer and paste there exactly this:
CREATE OR REPLACE PACKAGE AS TEST_PACKAGE
FUNCTION TEST RETURN BOOLEAN;
END TEST_PACKAGE;Try to compile it. You get nothing as compilation output. Yes in this case is part of first line underlined indicating there is some problem...
3) Correct the mistake and use this as package spec
CREATE OR REPLACE PACKAGE TEST_PACKAGE AS
FUNCTION TEST RETURN BOOLEAN;
END TEST_PACKAGE;4) Compile it -> You get "Compiled" as a message from the compilator
5) Create new empty file TEST_PACKAGE.pkb
6) Open it in SQL Developer and paste there exactly this:
CREATE OR REPLACE PACKAGE BODY AS TEST_PACKAGE
FUNCTION TEST RETURN BOOLEAN IS
BEGIN
  NULL;
END;
END TEST_PACKAGE;Try to compile it. Nothing happens. No compilator output at all... Why???
In this case even first line isn't underlined, but part of the function declaration is.
So the point is, when you do mistake in package declaration like I did here (I moved keyword "AS" before name of package) you don't get any message from the compilator.

Similar Messages

  • Worksheet parser bug

    Hi,
    I bet this has to be known bug, but I didn't find it here on the forum...
    Open some connection and type this in the worksheet:
    BEGIN
    DBMS_OUTPUT.PUT_LINE('abc');
    END;
    SELECT *
    FROM EMP;Now put the cursor either inside the anonymous block or on the query and pres CTRL + Enter.
    SQL Developer doesn't respect the end of the anonymous block and tries to execute the whole code as one script...

    Its not. In fact, it was engineered to be like SQL*Plus,
    If you try it in SQL*Plus, you will get the exact same issue. You need to add a / after your anonymous block for this to work
    HR@ORCL> !cat ttt.sql
    set serveroutput on
    begin
    dbms_output.put_line('abc');
    end;
    select 1 from dual;
    HR@ORCL> @ttt
    7 /
    select 1 from dual;
    ERROR at line 5:
    ORA-06550: line 5, column 1:
    PLS-00103: Encountered the symbol "SELECT"
    HR@ORCL>
    Decorating with the / gives you this.
    HR@ORCL> @ttt
    abc
    PL/SQL procedure successfully completed.
    1
    1
    HR@ORCL>

  • 3.1EA3 Code parser bug + File - New...

    Hello guys,
    I was creating new package yesterday, because it was just some testing package, I didn't need to store the source code in a file. Keeping it in database only was enough for me...
    So I used File -> New and chose Package. Ok, this created a package specification template. But how to create package body? There is nothing like that in the File -> New window. So this is first thing I wanted to mention.
    So I created empty file, opened it in SQL Developer with current connection and started typing package body. The source code looked like this:
    CREATE OR REPLACE PACKAGE BODY AS TEST_PACKAGE2
    END TEST_PACKAGE2;Yes, there is a bug in this code "AS" should be on the end...
    I wrote planty of code in the package body and when I finished, saved it and wanted to compile, nothing happened. The only thing what will happen after you try to compile this, is that the file name in the tab is changed to italic, like when you do some change.
    Why is the filename on the tab changing to italic?
    Why compilator doesn't report any error message in this case?
    Edited by: xxsawer on 11.1.2012 23:44

    Hello Vadim,
    I think you are missing the point. The point is that you make a mistake in declaring new package body or spec and when trying to compile it, you don't get any warning or error. The procedure how you create new spec is irelevant...
    So once again.
    1) Create new empty file with name TEST_PACKAGE.pks
    2) Open it in SQL Developer and paste there exactly this:
    CREATE OR REPLACE PACKAGE AS TEST_PACKAGE
    FUNCTION TEST RETURN BOOLEAN;
    END TEST_PACKAGE;Try to compile it. You get nothing as compilation output. Yes in this case is part of first line underlined indicating there is some problem...
    3) Correct the mistake and use this as package spec
    CREATE OR REPLACE PACKAGE TEST_PACKAGE AS
    FUNCTION TEST RETURN BOOLEAN;
    END TEST_PACKAGE;4) Compile it -> You get "Compiled" as a message from the compilator
    5) Create new empty file TEST_PACKAGE.pkb
    6) Open it in SQL Developer and paste there exactly this:
    CREATE OR REPLACE PACKAGE BODY AS TEST_PACKAGE
    FUNCTION TEST RETURN BOOLEAN IS
    BEGIN
      NULL;
    END;
    END TEST_PACKAGE;Try to compile it. Nothing happens. No compilator output at all... Why???
    In this case even first line isn't underlined, but part of the function declaration is.
    So the point is, when you do mistake in package declaration like I did here (I moved keyword "AS" before name of package) you don't get any message from the compilator.

  • Parser Bug: Bogus Extra XMLDecl

    Parser Bug: Bogus Extra XMLDecl
    I am finding that the latest xmlparserv2 is adding a spurious XMLDecl, which isn't even well-former XML. E.g., here's the output of a test program:
    Parsing:
    <?xml version="1.0" encoding="UTF-8"?><trivial/>
    Result:
    <?xml version="1.0" encoding="UTF-8"?>
    <?xml version = '1.0' encoding = 'UTF-8'?>
    <trivial/>
    Is this a known bug? Are there simple work-arounds? Will it be fixed?
    Here's the source, which I compiled with a classpath containing only xmlparserv2.jar from XDK 9.2.0.5_01 (dated 3/27/03, size 660,606 bytes).
    import java.io.StringReader;
    import javax.xml.transform.Transformer;
    import javax.xml.transform.TransformerFactory;
    import javax.xml.transform.dom.DOMSource;
    import javax.xml.transform.stream.StreamResult;
    import oracle.xml.parser.v2.DOMParser;
    import oracle.xml.parser.v2.XMLNode;
    import org.w3c.dom.Document;
    import org.w3c.dom.Node;
    import org.w3c.dom.traversal.NodeFilter;
    import org.w3c.dom.traversal.TreeWalker;
    public class TestOracleParser {
    public static void main(String args[]) throws Exception {
    new TestOracleParser().testOracleParser();
    public void testOracleParser() throws Exception {
    final String SIMPLE_XML = "<?xml version=\"1.0\" encoding=\"UTF-8\"?><trivial/>";
    System.out.println("Parsing:");
    System.out.println(SIMPLE_XML);
    DOMParser parser = new DOMParser();
    parser.setErrorStream(System.err);
    //parser.setValidationMode(DOMParser.DTD_VALIDATION);
    parser.showWarnings(true);
    parser.parse(new StringReader(SIMPLE_XML));
         Document document = parser.getDocument();
    DOMSource source = new DOMSource(document);
    StreamResult result = new StreamResult(System.out);
    TransformerFactory transFactory = TransformerFactory.newInstance();
    Transformer transformer = transFactory.newTransformer();
    System.out.println("Result:");
    transformer.transform(source, result);

    Your test appears to work fine with XDK 9.2.0.6.0.

  • 4500X command parser bug for monitor capture

    I have had a question about wireshark support on the 4500X switches. I thought there were some capabilities missing from my 4500X-24 running 3.5.2E but as I was entering my question here I realized what the problem is.  I thought capabilities were missing because the question mark doesn't doesn't display all the options:
    Switch#show monitor capture ?
      file  Display captured file
    Switch#show monitor capture mycap ?
    % Unrecognized command
    But it just seems to be a parser bug because it does actually accept and execute commands as documented:
    Switch#show monitor capture mycap parameter
       monitor capture mycap interface Vlan11 both
       monitor capture mycap interface TenGigabitEthernet1/1/7 in
       monitor capture mycap match ipv4  any any
       monitor capture mycap file location bootflash:mycap.pcap buffer-size 1
       monitor capture mycap limit packets 100 duration 60
    I just thought I would pass this along.
    -Jeff

    What was the issue and how did you solve it?
    -Deepak

  • [svn:fx-trunk] 11601: Integrating Min' s recent fixes to FXGUtils for a group of attribute parsing bugs and clarification of error messages .

    Revision: 11601
    Author:   [email protected]
    Date:     2009-11-09 22:00:50 -0800 (Mon, 09 Nov 2009)
    Log Message:
    Integrating Min's recent fixes to FXGUtils for a group of attribute parsing bugs and clarification of error messages.
    QE notes: Please confirm test cases.
    Doc notes: N/A
    Bugs:
    SDK-22982 - FXG compiler fails on percentage lineheight inside formatted text content
    SDK-24093 - If a
    tag has children, you do not get a compile error.
    SDK-24092 - You get a compile error if you set justificationStyle="prioritizeLeastAdjustment"
    SDK-24097 - Setting an invalid verticalAlign style on RichText has a typo in the error message
    Reviewer: Min
    Tests run: Checkintests, Bug test cases
    Is noteworthy for integration: No
    Ticket Links:
        http://bugs.adobe.com/jira/browse/SDK-22982
        http://bugs.adobe.com/jira/browse/SDK-24093
        http://bugs.adobe.com/jira/browse/SDK-24092
        http://bugs.adobe.com/jira/browse/SDK-24097
    Modified Paths:
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/fxg/FXGException_en.properties
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/RichTextNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/AbstractRich BlockTextNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/AbstractRich TextLeafNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/AbstractRich TextNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/BRNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/ImgNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/TabNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/richtext/TextHelper.j ava
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/dom/text/AbstractCharacte rTextNode.java
        flex/sdk/trunk/modules/fxgutils/src/java/com/adobe/internal/fxg/sax/FXGSAXScanner.java

    Dear Pallavi,
    Very useful post!
    I am looking for similar accelerators for
    Software Inventory Accelerator
    Hardware Inventory Accelerator
    Interfaces Inventory
    Customization Assessment Accelerator
    Sizing Tool
    Which helps us to come up with the relevant Bill of Matetials for every area mentioned above, and the ones which I dont know...
    Request help on such accelerators... Any clues?
    Any reply, help is highly appreciated.
    Regards
    Manish Madhav

  • SAX Parser Bug? Help!

    Javadoc for the SAX 1.0 API states its should be possible to reuse a parser object to parse one document after another. I have not been able to do this - I keep getting and error "EOF expected" even though endDocument() in HandlerBase has been called.
    Is this a bug?
    Thanks much...

    <BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by Joshua Introne ([email protected]):
    Javadoc for the SAX 1.0 API states its should be possible to reuse a parser object to parse one document after another. I have not been able to do this - I keep getting and error "EOF expected" even though endDocument() in HandlerBase has been called.
    Is this a bug?
    Thanks much...<HR></BLOCKQUOTE>
    Sorry - this isn't quite right. The final tag is parsed, but endDocument is never called. How can I get that event to be thrown without closing the stream feeding the inputSource?

  • OracleCommand Parsing Bug

    I am using Oracle.DataAccess.dll v2.111.6.0 and have encountered a bug in OracleCommand's parsing algorithm.
    This code should run properly:
    using (var con = new Oracle.DataAccess.Client.OracleConnection(connString))
    con.Open();
    using (var c = con.CreateCommand())
    c.CommandText = "SELECT 1 FROM dual WHERE ''=/**/'' and 1 = :p1";
    c.BindByName = true;
    var p = c.CreateParameter();
    p.ParameterName = ":p1";
    p.Value = 1;
    c.Parameters.Add(p);
    var r = c.ExecuteScalar();
    Instead it throws an OracleException with the message "ORA-01008: not all variables bound".
    With a little digging I found that the problem is that once the command parser in the OracleCommand class encounters the combination of characters - */' - it enters an invalid state which causes it to ignore any parameters defined afterwards and as a result the parameters are not sent to the oracle engine.
    Is this bug known and, if so, is there a fix?

    Hi Anthony,
    I saw you're using Windows XP, so you can use the "patch" utility found in the GNU tools for Win32. You can download the "patch" tool here: http://sourceforge.net/project/showfiles.php?group_id=23617&package_id=32165
    Download the patch that fixes the issue, http://www.oracle.com/technology/products/berkeley-db/xml/update/2.3.10/patch.2.3.10.2
    The patch tool should be run like this to apply the patch:
    BerkeleyDBXML_2.3.10_root_dir>"C:\Program Files\GnuWin32\bin\patch.exe" -p0 -i patch.2.3.10.2 --binary
    Regards,
    Andrei

  • NumberFormat parse() bug ? Please Help

    It seems like parse() method of NumberFormat fails silently for string with a number in scientific notation such as "1.456000E+05" -- I had to use a doubleValue(0 from Double().
    This very short piece illustrates it -- Can anyone please comment on any fallacies here ?
    Thanks.
    // Numberformat parse() fails when parsing
    // string to get double value
    // But Double.doubleValue() works
    import java.io.*;
    import java.text.*;
    public class jf {
    public static void main(String args[]) {
    String str = "1.456000E+05";
    System.out.println("Original string: str="+str);
    NumberFormat nf = NumberFormat.getInstance();
    try{
    double val = nf.parse(str).doubleValue();
    System.out.println("NumberFormat.parse() gives: val="+val);
    } catch (Exception e) {;}
    Double Val = new Double(str);
    System.out.println("Double.doubleValue(): val="+Val.doubleValue());

    Thanks, but changing to DecimalFormat ...
    import java.io.*;
    import java.text.*;
    public class jf2 {
       public static void main(String args[]) {
          String str = "1.456000E+05"; // 145600.00 in E format
          System.out.println("Original string: str="+str);
          DecimalFormat df = new DecimalFormat();
          try{
             double val = df.parse(str,(new ParsePosition(0))).doubleValue();
             System.out.println("DecimalFormat.parse() gives: val="+val);
          } catch (Exception e) {;}
          Double Val = new Double(str);
          System.out.println("Double.doubleValue(): val="+Val.doubleValue());
    }Still gave:
    Original string: str=1.456000E+05
    DecimalFormat.parse() gives: val=1.456
    Double.doubleValue(): val=145600.0Does this indicate a bug ? jre is 1.3.1

  • PL/SQL Parser Bug?

    Hi,
    just came across a strange PL/SQL parser behavior. Why is the following PL/SQL code valid?
    DECLARE
        vEmpNo NUMBER;
    BEGIN
        FOR rEMP IN
          ( SELECT EMPNO
              INTO vEmpNo
              FROM EMP
        LOOP
            dbms_output.put_line('Value: '||vEmpNo);
        END LOOP;
    END;The "INTO vEmpno" doesn't make much sense, but it's successfully compiled and executed.
    Database version is 10.2.0.3.0
    Any idea?
    Patrick
    My APEX Blog: http://www.inside-oracle-apex.com/
    The APEX Builder Plugin: http://builderplugin.oracleapex.info/
    The ApexLib Framework: http://apexlib.sourceforge.net/

    the pl/sql manual gives the syntax of a cursor for loop as
    cursor_for_loop_statement ::=
    [<< label_name >>] FOR record_name IN
    {cursor_name [(cursor_parameter_name
                  [, cursor_parameter_name]...,)]
    | (select_statement)}
    LOOP statement [statement]...
    END LOOP [label_name];and the pl/sql manual defines a "select into" as
    "The SELECT INTO statement retrieves data from one or more database tables, and assigns the selected values to variables or collections. For a full description of the SELECT SQL statement, see Oracle Database SQL Reference."
    and it also says
    "Selecting At Most One Row: SELECT INTO Statement
    If you expect a query to only return one row, you can write a regular SQL SELECT statement with an additional INTO clause specifying the PL/SQL variable to hold the result."
    so, if pl/sql feels that the INTO is just another clause to a regular select statement, and cursor-for-loops work on select statements, then there is no problem. huh?????
    guess in a similar way to v8 pl/sql allowing select of booleans, it just ignores that clause. but since it isn't explicitly stated as being allowed, we can hope that oracle will eventually notice and get rid of it.

  • SQL Parsing Bug in Oracle SQL Analyze in 9i

    The following does not work in SQL Analyze on 9i:
    select extract(HOUR from to_TIMESTAMP('2003-02-28 13:00:00','YYYY-MM-DD HH24:MI:SS')) FROM DUAL;
    It complains at there being two "from" in the SQL.
    As I'm new to OTN is there a fix for this bug and is it listed in a bug tracker?
    Thanks

    Hi Rodrigo,
    Thank you for reporting this. The only bug I see currently logged on a comment affecting the generated View DDL involves ending the last line of the definition with a comment, which treats the ending semi-colon (even if on a different line) as part of the comment. That is actually related to a low-priority bug against an Oracle database API.
    Unfortunately, your pseudo-code is a bit complex. Trying a few quick, simpler tests against the standard HR schema did not reproduce the issue. I tried INNER JOIN, and nested SELECTs. It would help greatly if you could provide a test case compilable against one of the standard schema, like HR or SCOTT.
    Regards,
    Gary
    SQL Developer Team

  • DefineFont2/3 tag parsing bug

    mxmlc report a "Unable to create source" error when I tried to<br />embed a swf file into flex application.<br /><br />Following is the stacktrace for this error.<br /><br />java.lang.StringIndexOutOfBoundsException: String index out of range: -1<br />     at java.lang.String.checkBounds(String.java:405)<br />     at java.lang.String.<init>(String.java:450)<br />     at flash.swf.SwfDecoder.readLengthString(SwfDecoder.java:275)<br />     at flash.swf.TagDecoder.decodeDefineFont2And3(TagDecoder.java:770)<br />     at flash.swf.TagDecoder.decodeDefineFont2(TagDecoder.java:737)<br />     ...<br /><br />I think this problem because of swfdecoder parsing font name wrong way.<br /><br />DefineFontInfo tag description in swf file format spec says,<br />  Note that font name strings in the DefineFontInfo tag are not null-terminated;<br />  instead their length is specified by the FontNameLen field.<br /><br />But SwfDecoder#readLengthString() try to remove null-terminator.<br /><br />Attached patch fix this problem.<br />Please entertain the patch.<br /><br />---<br />Sorry about my poor English.<br /><br />INADA Naoki  <[email protected]><br />   K Laboratory, KLab Inc.

    <DIV dir=ltr align=left><SPAN class=126172614-13052008><FONT face=Arial <br />color=#0000ff size=2>The code explicitly comments that removal of the null <br />terminator needs to be done, but as you said, this seems to contradict the <br />specification for fontName in DefineFontInfo or DefineFont2/3. Thankfully, <br />it seems readLengthString() is only used for font names.</FONT></SPAN></DIV><br /><DIV dir=ltr align=left><SPAN class=126172614-13052008><FONT face=Arial <br />color=#0000ff size=2></FONT></SPAN> </DIV><br /><DIV dir=ltr align=left><SPAN class=126172614-13052008><FONT face=Arial <br />color=#0000ff size=2>Before we change the code we'll ask the Flash team to <br />confirm whether the spec or our code is wrong. </FONT></SPAN><SPAN <br />class=126172614-13052008><FONT face=Arial color=#0000ff size=2>(However, from <br />the error message reported, it seems that an empty string (0 length) was <br />encountered for the font name and that seems invalid to <br />me...).</FONT></SPAN></DIV><br /><DIV dir=ltr align=left><SPAN class=126172614-13052008><FONT face=Arial <br />color=#0000ff size=2></FONT></SPAN> </DIV><br /><DIV dir=ltr align=left><SPAN class=126172614-13052008><FONT face=Arial <br />color=#0000ff size=2>Pete</FONT></SPAN></DIV>

  • Buggy(?) sql worksheet parsing

    When I have more statements in SQL worksheet I can run one of them by [F9] - run current one. But when I try it with these 2 statements it will fail with ORA-911 Invalid character.
    When I select 1st or 2nd one with 'text editor selection', [F9] works fine.
    select product_code, decode(active_to, null, '-', 'Y') active, count(1)
    from product_instances
    where product_code like '%MySMS%'
    group by product_code, decode(active_to, null, '-', 'Y') order by 1;
    select count(1) from PRODUCT_INSTANCES;
    any thoughts?

    The problem seems to be related to identifying the
    start and finish of the statement.
    A work around is to manually select the statement
    that you want to run (without the ;) and then use F9
    - I tried this with the first query in the example
    here and I got past the ORA-911 (got an ORA-942 as I
    don't have the right tables:)).Yes, thats another way to run a statement, in fact its also used for running sub selects in the middle of larger statements.
    Some things from the Raptor perspective that would
    help are:
    - highlighting the "current" statement it is
    s attempting to be runWhy would we do this? Clarity perhaps, The caret is on the line of the statement.
    - jumping to the syntax error location (this has
    s been discussed in a number of other threads,
    including several workarounds for the fact that JDBC
    does not support this natively)We have added a partial solution for scripts that will tell us where the error is and what it was, but we can only get the line number of the start of the statement. Only with anonymous blocks, can we figure out the relative line no and column number of the error.
    It is just my impression (I am being lazy and not
    going back and checking), but most cases where I have
    had the problem or that have been given as examples
    in the posts here seem to have either comments
    (prefixed by --) or %s in strings.

  • Xquery dateTime parsing bug

    I ran across a parsing error in dbxml shell while running an XQuery. You can replicate it by the following (BDBXML 2.3.10 on Windows XP):
    // Note the seconds in dateTime is 59.00
    dbxml> query 'let $dt := xs:dateTime("2008-12-30T22:59:59.00Z") for $a in $dt re
    turn $a'
    1 objects returned for eager expression 'let $dt := xs:dateTime("2008-12-30T22:5
    9:59.00Z") for $a in $dt return $a'
    // Note the seconds in dateTime is now 59.01
    dbxml> query 'let $dt := xs:dateTime("2008-12-30T22:59:59.01Z") for $a in $dt re
    turn $a'
    stdin:352: query failed, Error: Invalid representation of dateTime [err:FORG0001
    ], <query>:1:12
    let $dt := xs:dateTime("2008-12-30T22:59:59.10Z") for $a in $dt return $a
    ^
    This happens for seconds with a fraction greater than zero. Furthermore, if the minute value is less than 59, the error is eliminated:
    dbxml> query 'let $dt := xs:dateTime("2008-12-30T22:59:58.01Z") for $a in $dt re
    turn $a'
    1 objects returned for eager expression 'let $dt := xs:dateTime("2008-12-30T22:5
    9:58.01Z") for $a in $dt return $a'
    Is this an xqilla10 issue?

    Hi Anthony,
    I saw you're using Windows XP, so you can use the "patch" utility found in the GNU tools for Win32. You can download the "patch" tool here: http://sourceforge.net/project/showfiles.php?group_id=23617&package_id=32165
    Download the patch that fixes the issue, http://www.oracle.com/technology/products/berkeley-db/xml/update/2.3.10/patch.2.3.10.2
    The patch tool should be run like this to apply the patch:
    BerkeleyDBXML_2.3.10_root_dir>"C:\Program Files\GnuWin32\bin\patch.exe" -p0 -i patch.2.3.10.2 --binary
    Regards,
    Andrei

  • SQL parser bug?

    Is there a known problem with NOT IN operator? The following statement returns "no rows selected" message
    SELECT col_a FROM tab_a WHERE col_a NOT IN (
    SELECT col_b FROM tab_b);
    while the following two statements both return the expected result
    SELECT col_a FROM tab_a MINUS SELECT col_b FROM tab_b;
    SELECT col_a FROM tab_a WHERE NOT EXISTS (
    SELECT * FROM tab_b WHERE col_b = tab_a.col_a);
    Note 1: This failure only happens on certain tables, but not on other tables in the same database.
    Note 2: both of tab_a.col_a and tab_b.col_b are VARCHAR2(32).
    Note 3: Oracle8.1.7 on Tru64.
    Any insights are appreciated.
    Robin

    My first instinct is that it has to do with handling nulls. Each of those statements will do it slightly differently.
    col_a NOT IN () : if col_a is null won't return row
    MINUS : Oracle will correctly compare nulls
    NOT EXISTS ( = col_a) : if col_a is null won't return row so 'not exists' will be true and master query row will always return row (even if they were both null).
    Have you checked that?
    Richard

Maybe you are looking for

  • Moderator please – I want the money back that BT o...

    Could a BT Forum Moderator please look into my account, contact the department currently handling my case, and arrange a refund for the money that BT owe me back, after overcharging my account for the last 3 months. After my original phonecall 7 WEEK

  • My "edit playlist" button has disappeared!

    Recently, my "edit playlist" button has disappeared all together when I right-click a playlist. I do know how to manually add to a playlist, but it's inconvenient. It's there when I click File, but I can't click on it. I've tried calling Apple, but o

  • Importing swfs question

    Hi I have a swf ("A") that imports two other swfs ("B and C").  This works fine. I have another swf ("D") that imports swf "A".   D imports swf A, but suddenly swf A won't import B and C - a URL error gets thrown. I have played with the publish setti

  • Send streaming password with GET or POST

    hi! we are using a plugin for protools which outputs a quicktime stream for our clients during recording sessions. this stream is password protected. since the client hearing/viewing this stream is already on a password protected site it would be nic

  • IPod update timing out?!

    For the last week since the iOS 4.2 update came out, I've tried downloading it with no success; the file does not process or verify, the network connection times out (Error code is -3259) when the file is totally completed, according to iTunes (missi