Looks like a bug in DB, Getting a invalid number with a group by

Hi All,
This is very strange problem that I am encountering. Would like to hear from you, if you have also witnessed something similar:
What I am doing:
basically I am doing something like:
Select x,y,z,count(distinct a) from
taba,tabb,(select col1,col2 from tabc) tc,
tabd left outer join tabe on tabd.col=tabe.col,
viewf
where (joins between the tabs)
group by x,y,z
All the join columns are number(10,0) in all the tables/views.
If I run this query I get an invalid number 01722. If I comment the count(distinct a) column, the query runs fine.
Or if I put the to_number for all the joined columns in the where clause, the query runs fine.
I have never seen such behaviour of Oracle database. May be I am doing something really wrong?
Looking for help.
Thanks in advance.

Well, I do not think it is related to GROUP BY or to DISTINCT. Most likely you have some column X that holds both numeric and non-numeric strings and some other column that indicates if column X value is numeric. If this is the case, you can get ORA-01722 simply becuase Oracle does not apply where clause predicates in the order they are listed. Assume column c1 holds numeric strings when column c2 is set to some specific value. Even if you use:
WHERE c2 = 'some-value'
AND TO_NUMBER(c1) = 5
it is not guaranteed Oracle will first check c2 and only then c1. And if Oracle decides to check c1 first, TO_NUMBER will fail on the first non-numeric c1. Especially it can happen in not very recent Oracle versions. In recent versions most of the time optimizer is smart enough to realize that checking c2 costs less that converting c1 to number and then checking it, so it will check c2 first. Anyway, you need to use ORDERED_PREDICATES hint to let optimizer know to follow predicate sequence. For example, in 9i:
SQL> create table tbx(flag varchar2(6),value varchar2(10))
  2  /
Table created.
SQL> insert into tbx values('alpha','abc');
1 row created.
SQL> insert into tbx values('number','123');
1 row created.
SQL> commit;
Commit complete.
SQL> select * from tbx
  2  where flag = 'number' and value = 123
  3  /
where flag = 'number' and value = 123
ERROR at line 2:
ORA-01722: invalid number
SQL> select /*+ ORDERED_PREDICATES */ * from tbx
  2  where flag = 'number' and value = 123
  3  /
FLAG   VALUE
number 123
SQL>SY.

Similar Messages

  • It looks like I currently share an apple id and account with my daughter.  Our phones are synched even alarms go off on each others phones.  I want to have my own account so this doesnt continue to happen.  How do I change my iphone to have its own id?

    It looks like I currently share an apple id and account with my daughter between iphones.  When I set my alarm it goes off on both phones and contacts are synched as well.  How do I seperate from  her account?

    According to Create an AppleID:
    Choose an Apple ID and password.
    Enter your primary email address as your Apple ID. This will be used as the contact email address for your account. Please note that this email address must be verified before you can use certain Apple services.
    And, unless I'm missing something, it appears that you're trying to use your father's e-mail address. Start at Find your AppleID and see if you can resolve the issue.

  • How can i get kinect serial number with KinectSDK1.8

    Now I'm getting depth datas and color datas from 4 Kinect with Kinect SDK 1.8.
    I use Windows 8, Visual Studio 2013, Kinect SDK 1.8 and Kinect v1 × 4
    /***** getINuiSensor Instance *****/
    INuiSensor* sensor;
    HRESULT result = S_OK; // device_id : 0~3
    result = NuiCreateSensorByIndex((int)std::stoi(device_id),&sensor);
    But this way, the couple of the Kinect and "device_id" 
    such as the order of the device connection is changed.
    Therefore, in order to specify the device to start more explicitly,
    I would like to know how to get the serial number of kinect on Windows.
    I know the way that There is a way to get INuiSensor Instance 
    by a serial number getting from the device manager with OpenNI.
    With Kinect SDK, how to get the INuiSensor Instance using the serial number ?
    If Kinect SDK have no way it, please tell me some good ideas about that.

    The serial number is printed on the sensor. We do not provide an api that gives you the serial number of the device.
    As for the api you are using, that will only provide a unique identifier for the port the device is connected to. Plugging in another Kinect device to the same USB2 port will get you the same value. See other threads that discuss this.
    There is nothing at a Kinect SDK level that will allow you to identify the exact device. You may need to look at Win32/OS level api's to see what is exposed through the raw driver. A tool to query the device that will let you see what these api's can provide
    is USBView.
    Carmine Sirignano - MSFT

  • I have bought Adobe Photoshop Elements 13 and I try to get my serial number with my activation ID but I get a message that my ID is not good. What can I do?

    Where can I get a serial number with my activation ID if it dosen't work on the ADOBE site?

    Well, for your ID just use the ID and password you're using here. If you meant your redemption code rather than your ID, try this:
    Redemption Code Help

  • Looks like a bug in Date class

    Hi guys,
    I just logged in to highlight what I encountered when I was
    trying to convert milliseconds of a time stamp to a date/time
    string.
    Just see the following codes.
    public function
    convertTimeStampToDateAndTime(timeStampInMilliseconds:Number):String
    var stampReadableFormate:String = null;
    if(!isNaN(timeStampInMilliseconds))
    var stampDate:Date = new Date(timeStampInMilliseconds);
    var stampHour:Number = stampDate.getHours();
    var stampMin:Number = stampDate.getMinutes();
    var stampSec:Number = stampDate.getSeconds();
    var stampYear:Number = stampDate.getFullYear();
    var stampMonth:Number = stampDate.getMonth() + 1;
    var stampDate:Number = stampDate.getDate();
    stampReadableFormate = "" + stampDate + "/" + stampMonth +
    "/" + stampYear + " " + stampHour + ":" + stampMin + ":" +
    stampSec;
    trace("Date/Time=" + stampReadableFormate);
    return stampReadableFormate;
    Then I call this method from somewhere by using,
    var reportTime:String =
    this.convertTimeStampToDateAndTime(new Date().getTime());
    You can see trace method in convertTimeStampToDateAndTime()
    is giving you a correct output.
    But just change the convertTimeStampToDateAndTime() method in
    a way that you get time information after you got date, month and
    Year. To explain what I am taking about I have pasted changed code
    as following.
    public function
    convertTimeStampToDateAndTime(timeStampInMilliseconds:Number):String
    var stampReadableFormate:String = null;
    if(!isNaN(timeStampInMilliseconds))
    var stampDate:Date = new Date(timeStampInMilliseconds);
    var stampYear:Number = stampDate.getFullYear();
    var stampMonth:Number = stampDate.getMonth() + 1;
    var stampDate:Number = stampDate.getDate();
    var stampHour:Number = stampDate.getHours();
    var stampMin:Number = stampDate.getMinutes();
    var stampSec:Number = stampDate.getSeconds();
    stampReadableFormate = "" + stampDate + "/" + stampMonth +
    "/" + stampYear + " " + stampHour + ":" + stampMin + ":" +
    stampSec;
    trace("Date/Time=" + stampReadableFormate);
    return stampReadableFormate;
    now call
    var reportTime:String =
    this.convertTimeStampToDateAndTime(new Date().getTime());
    Now see the trace out put, what has happened, your timing
    informations are undefined.
    Looks like it is a bug to me.
    If this issue has already been discussed before I am very
    sorry by duplicating it.
    Chears
    Naleen

    Thanks for all the responses.
    I am not redefining stampDate inside your function/method.
    Please look at it again.
    I call convertTimeStampToDateAndTime() from out side and in
    my original post I have written it as following....
    var reportTime:String =
    this.convertTimeStampToDateAndTime(new Date().getTime());
    That is just an example for you to udnerstand it,
    But what we are doing in our actual application is we read an
    XML file which has time stamps on each elements as attributes. For
    an example, if following is the XML line....
    <element timeStamp='1195770453174'
    >blaaa</element>
    In Actionscript we call this method,
    var reportTime:String =
    this.convertTimeStampToDateAndTime(1195770453174);
    So I am not redefining at all.
    Just do a simple test by user self.
    My second point is.... If I am redefining a date again with
    same number of Milliseconds Date class constructor shouldn't care
    it.
    For an instance
    Var myMilliSEC1:Number = new Date().getTime(); // this is
    constructing today's date and converts to milliseconds
    var myDate1:Date = new Date(myMilliSEC1); //so this should
    give me the todays date again...
    Var myMilliSEC2:Number = myDate1.getTime();
    Now myMilliSEC1 should equal to myMilliSEC2
    But that is a different issue.... I am just answering for
    your comment on "REDEFINING DATE":
    Anyway thanks for your time and concern.

  • HT1766 Can I pleas go back to the earlyer iOS on my iPhone? I hate ios7; if I wanted something that looks like windows 8 I would get a PC. Is ther a way to stop the promp to conver on my iPad? I want nothing to do with windows or cartoons. I love Apple sy

    I want to remove ios7 on my iPhone 5. It was a mistake to try to look like windows8, in my opinion. I naught app PC for my church; it has Windows 8, and we all hate it. It is pale and cartoonish, my wife can't even see how to unlock my phone by spelling my pets nick name. It has been my experience that people by Apple to get away from Window and Android. Speaking for all the people I know with Apple products. Yes!!! We still think you are pretty.
    I want iOS 6.1.4  back

    No.
    Apple showed off iOS 7 at a special event, put up a web page.
    The new look was not a surprise.
    Why did you update if you did not like it?

  • I just bought an ipod touch 32gig second hand and i would like to know how to get it to sync with my i tunes content

    how do i get up and running with a recently purchased 5th generation ipod touch 32 gigs  that i bought second hand?

    What happens when you sync it:
    iTunes: Syncing media content to iOS devices and iPod
    If the iPod has content for the previous owner, you should erase the iPod and setup the iPod with your ID and media
    Go to Settings>General>Reset>Erase all content and settings

  • Why do I get Error invalid number when I try to send a text to my husband?

    I tried to send my husband a text this afternoon from my phone. I keep getting a message from +1 (1) (216) 116-11. It says the following:
    7657905 Error Invalid Number
    Please resend using a valid 10 digit mobile number or valid short code.  HELP!

    My wife had this problem for the past few days for a specific phone number. These instructions fixed it -- without having to restore from backup or contact the carrier.
    http://www.timothydeblock.com/blog/2014/5/12/fix-message-sent-using-invalid-numb er-of-digits-msg-2114
    Open Contacts -> select the contact -> select edit -> scroll all the way to the bottom and delete contact (delete multiple entries of the same phone number, my wife had five).
    Open the Messages app -> select Edit -> select the red circle and then select delete. Do this for both the person and the messages you received.
    Open the Settings app -> scroll down and select Messages -> turn off messages, by selecting the switch, and any other options turned on.
    Turn off your phone and then turn it back on.
    Go straight to messages and compose a new message and put in the phone number of the person you're trying to text. Send and that person should receive the text message. Add the person to your contacts and go back into message settings to turn on any other options you want on that you turned off.
    John

  • HT4623 Looks like a Bug in the Default Mail application of iOS 6 in an iPhone 4S

    In the mail application of my iPhone 4S running iOS6, while composing a new email, the recipients field marked as "To: " doesn't respond in a single tap rather requires two or three and sometimes even more taps for the blue input bar to appear and enable typing of the recipient's email address.
    I have tried resetting the iPhone completely to factory settings and erasing everything several times, but the problem still persists.
    It has nothing to do with the suggested automatic recipient addresses as the field doesn't gets activated itself in first tap and the blue bar doesn't appear unlike all other fields such as, "Subject", Cc, etc. It looks as if the touch screen didn't get any feedback and hence returned no response.
    Once after a few tapping trials, the blue input bar appears indicating the field ready for the recipient's address to be typed and then everything gets normal from there.
    Rest all the applications and everything on the iPhone is working perfectly fine with no problems anywhere at all.
    Kindly suggest the necessary measures.
    Also, I am facing exactly the similar situation with my iPod Touch 4th generation running on iOS 6 as well. This has made me conclude that it must be a bug in the mailing application.
    Both the devices are only a month old. Also they are configured for different account s and have completely different dat from one another as the ipod touch is being used by my sister who has a separate masbook to sync it.

    That connection is supported for tethering....
    Barry

  • Looks like a bug in Java ... what's going on he

    I can't really recreate this problem, but it occurs in situations like the one below. Unfortunately the program is just too complex to post it all here. In other words, I'm not sure if this exact code will produce the bug, but it happens sometimes in similar situations:
            System.out.println("arrived here");
            // while not ready
            System.out.print("waiting until initialisation is ready..."+AnimationFrameManager.printCurrentFrame()+" ... ");
            // System.out.println("THIS LINE HAS BEEN PRINTED");
           while (!this.isReady)
                // do nothing
            System.out.println("done!");The bug is that I get output like this:-
    arrived here
    waiting until initialisation is ready...3...waiting until initialisation is ready...3...done!Why is it printing twice?
    Weirder still is that if I put a line in-between the last println statement and the beginning of the while loop (for example, the line that is commented out above), then it works as expected!
    i.e.
    arrived here
    THIS LINE HAS BEEN PRINTED
    waiting until initialisation is ready...3...done!If I take the static method call out, it works as expected:
            System.out.println("arrived here");
            // while not ready
            System.out.print("waiting until initialisation is ready... "); // <!-- took the static method call out
            // System.out.println("THIS LINE HAS BEEN PRINTED");
           while (!this.isReady)
                // do nothing
            System.out.println("done!");Any ideas?
    Edited by: GallahJava on Feb 2, 2010 7:47 PM
    Edited by: GallahJava on Feb 2, 2010 7:52 PM

    There is no GUI, just console output.
    I realise there is multi-threading going on "behind the scenes", but I haven't explicitly created any threads except for one; this is the one-and-only thread created by me, does this help? I can't imagine how it could possibly be responsible for these "multithreading issues".
    public class AnimationRunner implements Runnable
        private SuperImage[] images;
        private int imageNo;
        private static final int MAX_IMAGES = 39; // How many images in the progression?
        private static final int SWAP_DELAY = 850;
        public AnimationRunner()
                images = new SuperImage[MAX_IMAGES];
                System.out.print("\nInitialising images to memory...");
                for (int x = 0; x < MAX_IMAGES; x++)
                    String imageFile = "images/anim-frame-" + (x + 1) + ".png";
                    //System.out.println(imageFile);
                    images[x] = SuperImage.createFrom(imageFile);
                System.out.println("done!");
                imageNo = 0;
                new Thread(this).start();
        public void run()
            System.out.println("Animation has started.");
            while (true)
                try
                    Thread.sleep(SWAP_DELAY);
                    if (imageNo >= MAX_IMAGES - 1)
                        imageNo = 1;
                    else
                        imageNo++;
                    //System.out.println("Now on image #" + (imageNo + 1) + "\n");
                catch (Exception e)
        public int getImageNo()
            return imageNo + 1;
        public SuperImage getImage()
            //System.out.println("\n(using image no."+(imageNo+1)+")\n");
            return images[imageNo];
    }

  • [SOLVED] Cannot boot (doesn't look like i915 bug)

    I can't boot my laptop.  Immediately after pressing enter in grub, I see black.  I put my laptop to sleep when it had about 10% battery remaining, and I'm guessing that it ran out of juice before I plugged it in.
    My last successful boot was just after installing linux 3.16.3-1 on September 21.  I booted with archboot and did a chroot.  The only "suspicious" packages I upgraded after this boot were xorg-server+common+xephyr and xf86-video-intel.  So I downgraded these.  No improvement.
    I read about the i915 kernel bug and couldn't find any lines mentioning i915_gem_stolen.c in journalctl -r.  But I downgraded my kernel to the kernel I had been using for a while, 3.15.5-2.  After this, I was happy to see the normal boot messages appearing.  However, when the console login appears, I can't input my user name.  Rather it disappears and reappears every second.  And, interestingly, it writes "Arch Linux 3.16.3-1-ARCH (tty1)" there instead of the kernel version I just downgraded to.
    The same behavior (disappearing and reappearing console login) happens when booting in recovery mode.
    Which logs do I need to look at in order to figure out what is going on?
    Thank you!
    Last edited by tinte (2014-09-27 15:41:01)

    Try rebooting with iso & doing the chroot thing then try to install the lts kernel & its headers/modules & then check everything.log etc for further clues & reboot after.
    The i915 bug has been a weird one with a whole host of different effects for everyone but as you are getting odd behaviour after a kernel downgrade I really would suggest the lts route just to see if that will give you a working system.
    good luck

  • Occasional invalid modified date on a file - looks like a bug

    In the Bridge script I'm working on, I need to compare the last modified dates of files. I've found that when I iterate through the app.document.visibleThumbnails array that in nearly every directory I test on, there are one or two files that return an "undefined" value for app.document.visibleThumbnail[i].spec.modified. It is not the same two files each run, but it's always at least one that has a missing value. It's a pretty simple test app to illustrate the problem:
    var thumbs = app.document.visibleThumbnails;
    for (var j in thumbs)
    var modDateSpec = thumbs[j].spec.modified;
    if (modDateSpec == undefined)
    Window.alert("modification date is undefined");
    Unless this value is not intended to be reliable, this appears to be a bug which is why I am reporting it. I have worked around the problem for now by getting the fsName from the same thumbnail, creating a new File object with that fsName and getting the modified date from that temp object.
    I've verified that purging and rebuilding the cache does not make the problem go away. I see this problem on many different directories.
    Is this the right place to report this type of bug or is there a different procedure I should follow?
    --John

    pessex wrote: Anybody know?  Or know of a different way that I COULD do it in Terminal?
    Don't know about Terminal, but, because you are using 10.9.x and iMovie, try the iMovie method:

  • Acrobat  CD Form Button Display incorrectly  when an effect  or transparency  is applied  Via indesign  Looks like a bug

    I have tested existing interactive pdfs that function perfectly in Acrobat XI and Adobe acrobat Reader but open the same files in Acrobat DC and Acrobat DC Reader and most of the documents i have designed and produces over last few years have been renderer useless
    Basically any interactive  form field wich uses  Effects,  Transparency  or  mode changes  like multiply  in every instance the  whole form element appear as a pixelated mess
    I have been posting  in an indesign Thread  https://forums.adobe.com/message/7427566#7427566  it contains  picture and a sample document
    Steve Werner has been helping  me   and has suggest  i post here too
    Graeme,
    I downloaded your PDF file. I have both Acrobat DC and Acrobat XI Pro and I can confirm that I view it the same way that you do.
    It clearly seems to be a bug. In the Acrobat beta group, most users are probably not InDesign users, and most of us who do use InDesign probably didn't create or test Show/Hide buttons using those forms of transparency. I'm sure it's not intentional, probably a side effect of some other change.
    It needs to be reported as a bug. I'd suggest using the Bug Report form here:
    Adobe - Feature Request/Bug Report Form
    Also, there are several Acrobat forums. I'd suggest posting (if you haven't already) in the these two forums:
    Rich Media & 3D
    PDF Forms

    It's good that you'll be submitting a bug. This sort of thing has happened before, and I think it's related some to the fact that InDesign is capable of generating field appearances that you can't set up in Acrobat (transparency in particular).

  • Extra bars in score set -- looks like a bug

    I've been working on some tunes and wanted to display both the bass track and the track with lyrics. Logic Pro 8 seems to have some trouble with this. A bug seems to add additional portions of a bar. Check out this screen shot and note the extra bar 21. If I look at individual tracks, it is fine. If I change windows and go back, it seems to refresh the screen correctly. This problem creates a big problem when I'm trying to play along.
    Screen shot:
    http://www.punkjob.com/ScoreSet.jpg

    you'd be far better off displaying this in a seperate editor the 'old' way, if you ask me. incidentally is this a LP7 song or freshly minted in 8?
    nice bug BTW.

  • Looks like a bug in RoboHelpX6 - CSS

    I have been working in RoboHelpX6 on some CSS - Initially I
    updated an existing CSS, and changed the font that was being used.
    Was using Tahoma and changed it to Verdana, when I did this the CSS
    lost the word wrap. I then went and created a new CSS from scratch,
    stated it out with Tahoma like the first one, everything worked
    fine. Went in and changed the font to Verdana and again lost the
    word wrapping!!!
    Any reports on this yet?

    RoboHelp x5 rewrote CSS as well. I'm pretty sure they think
    it's a feature.
    The workaround I developed involves authoring/editing the
    .css files outside RH, keeping them under version control, and
    reverting them whenever RH tries to change them, plus copying clean
    .css files over the rewritten ones after projects have been
    generated.
    Note that for every stylesheet fname.css you have, RH will
    create a supposed "Netscape" version of the stylesheet called
    fname_ns.css. Again, I'm pretty sure they think this is a feature.
    Unless you like the way the _ns.css version looks on
    Netscape/Firefox/whatever, you will want to copy your clean
    fname.css over fname_ns.css as well.
    Amy

Maybe you are looking for