A matter of design

Similar Questions

• A matter of design using the custom in custom frames fields.

  A matter of design, which is a good implementation of the following?

  I read the articles in the Knowledge Base 'Create custom fields' and 'Create a custom handler for a screen presentation' and what I want to do is control any arbitrary custom field - a Manager custom.

  For example, if I have the following fields:

  1. Field
  2. Field
  3. Field
  4. Field
  5. Field

  I want the second and fifth fields to have a different margin, filling or the width of the first, third and fourth fields, I would by methods layout() and object to the second and fifth fields - which shall replace the custom method of the sublayout() Manager logic?

  Or can I add logic in the custom handler, I find the fields with the method getField() to get the second and fifth fields and change the margin, filling or width with the logic in the custom handler?

  What is the right place to put the logic?

  Thank you.

  the field can only work with the space assigned by the Manager. how he uses this space is to the field, he could see itself smaller, for example.

• A matter of design about the update of a screen of classes different all in now the lock of the event.

  I have a screen:

  • A class that extends the screen

  I have three Classes that contain a call to a database that returns records of ~ 50 that I fill a ListField with. The ListField is added to a different Manager according to the class

  1. A class that contains a VerticalFieldManager
  2. A class that contains a HorizontalFieldManager
  3. Another class that contains a HorizontalFieldManager

  Once I complete the list and add what it the respective Manager, how can I add the Manager of one of the three classes of the class that extends the screen when I lock the event?

  Is it a bad design?

  There is a lot of code that will fill the lists, so I thought that using separate classes would be the way to go.

  I guess I can't understand how to add the handler to the class that extends the screen.

  If it is a bad design, which is a good approach?

  Yes, it should work. A couple of optimizations:

  (1) you don't have to check the isForeground() - even if your application is currently in the background, you want the user to see an update of the screen once it is implemented, not you?

  (2) you already have myScreen as the reference on the active screen - use instead of run this getActiveScreen() second time:

  myScreen.replace (oldVfm, value);

  I guess you can mark this as resolved issue.

• A matter of design of LAN switch

  Let's say in a configuration there are two standalone Cisco 6509 switch core (core 1 and core 2), connected each other with between etherchannel link. Interface VLAN (vlan 10,20,30,40) was created in the two switches to base with the HSRP configured in VLAN to provide gateway redundancy, currently interface gateway active all the VLANs are 1-based, and core 2 acts as a sleep state.

  There are also 6 network switches, each access switch with an uplink connected to the basic 1 and 2 respectively to provide basic redundancy. All switches access have all positions of vlan 10,20,30 and 40 associated and configured with them.

  Then, of course, that Cisco 6509 is the place of the idea to become the bridge root STP in the installer, I think that

  (1) core manually configured 1 as the brigdes root for all the VLANs

  core1 (config) # spanning - vlan 10-40 the square root of the drive shaft

  Core2 (config) # spanning - vlan 10-40 the square root of the secondary shaft

  or

  (2) manually configured and separate the brigdes root in core 1 & 2 core

  core1 (config) # spanning - vlan 10-20 root of the drive shaft

  core1 (config) # spanning - vlan 30-40 the square root of the secondary shaft

  Core2 (config) # spanning - vlan 30-40 the square root of the drive shaft

  Core2 (config) # spanning - vlan 10-20 root of the secondary shaft

  So who is the best choice and best practices in this case to get the best performance, or there is no significant different between them

  BTW, the Cisco 6509 core is and access switches are all performed on spanning tree pvst mode. Thanks in advance

  Best regards

  Wai

  To be honest, we had the same exact configuration with 60 switches to user access (each port a 48 and 98% of the capacity) and 63 VLANS and 6509 switch CPU was 1-2% with a SUP720. At peak hours I sometimes saw him go to 5%!

  We started to L2 and L3 load balancing but through the network with servers linux and gluing problem (dman server guys) came back to us to option 1. Never had a problem since.

  Then you should be OK.

  HTH,

  Ian

• Create a flat file reconciliation events - a matter of design

  Hello
  
  I am currently evaluating an existing implementation of IOM to rebuild using OIM 11 g and have questions about the ideal method to create events for the reconciliation of a flat file.
  
  The current implementation uses a web service call to process a flat file and creates reconciliation events. It runs every hour.
  
  Although this looks cool, but I thought there was no need to go as far.
  
  If IOM cannot consume the flat file directly, which means that if he needs massage data, I can always load the flat file data in an external table, write a pl/sql procedure to transform the data and put it into a global temporary table, and create events of reconciliation like that.
  
  What would be the ideal method to load data from a flat file in IOM?
  
  Thank you
  
  Khanh

  If it is a flat file, then watched GTC option? And why no staged between the two? IOM can read flat files very well either through GTC or write your own recon code.

  -Marie

• Confirming a design where I need to save where the user navigated to the screen.

  A matter of design, which is a good implementation of the following?

  I need to be able to save the location of the place where the user has scrolled, then when the user:

  1. Changing screens and returns to the previous screen.
  2. The application farm (moves the application at bottom) and opens the app again (move the app in the foreground).

  The user scrolls automatically here where the user was before.

  For example, if I have a VerticalFieldManager with the following text read only fields:

  • LabelField
  • LabelField
  • LabelField
  • LabelField
  • LabelField
  • ButtonField

  The user scroll down to the third LabelField — two actions user above performs, returns to the screen, and scrolls down to the third TextField again.

  Is the hardest part, I can't count on own dirty, muddy, or fields because it is read only the data and a ButtonField.

  I can't use Manager.setFocus because the fields have an accent.

  I think I can use Manager.setHorizontalScroll and Manager.getHorizontalScroll, but is there a better way?

  Not sure that your arguments to change my suggestion.

  "The user must be able to close the application."

  If they close the applciation that they lose their position.  It happens in most of the applications.  Browsers do not open up on top of the page you were viewing when you close, for example.

  "The user must be able to request the application thoroughly using the end key of red color"

  You will not be able to stop this.  It will just happen.  Nothing for you to do.

  "This is a pile of high display a screen where the Menu button is the main navigation.".

  This does not preclude that I proposed either possible.  If you pop a screen and then later push it, my understanding (and experience) are that he's going to be redisplayed as it was when you jump it.  The screens are objects, they don't stop screens just because they aren't on the stack of the screen.

  Also, I see no problem in what you descrtibed for a touchscreen device.

• How can I design a custom button that has three lines of text in the button itself?

  A matter of design, which is a good implementation of the following? I need to create a custom button that looks like:

   -----------
  | Header    |
  | Subheader |
  | Text      |
   -----------
  

  I know that I can not add a VerticalFieldManager or add any field of a ButtonField, but if I create a custom, button how I would create three lines of text similar to having three LabelFields lined up in a VerticalFieldManager?

  Thank you.

  You can create a focusable multiline field label creation with the bit style FOCUSABLE and with newline characters in its text. It is possible to work a bit like a button field (substitute navigationClick / navigationUnclick to perform the action or, better yet, to call fieldChangeNotify so that you can plug into the field with a FieldChangeListener).

  Use RichTextField rather than LabelField if you want to play with fonts (say, "BOLD" underlined header; italic subheader; normal text).

  If you want rounded corners and other effects, override the paintBackground method or painting of the field (don't forget to call the super.paint, if you go with the latter).

  That being said - that there is no single best design, so feel free to browse the forums (research of "multiline" or "multiline" yields quite a few results) and come up with your own ideas.

• A question about how to properly manage a single timer on a screen, when to schedule and cancel?

  A matter of design, which is a good implementation of the following?

  I have a single class that extends screen. In this class, I have a unique clock and the internal class that extends TimerTask.

  public class MyScreen extends MainScreen {
  
    private Timer _t;
  
    private class MyTask extends TimerTask{
      public void run() {
      ...impress your girlfriend with cool logic...
      }
    }
  }
  

  • What is the best place to set the timer to run the timer whenever the screen is loaded?
    1. Constructor
    2. ScreenUiEngineAttachedListener
  • What is the best place to cancel the timer to ensure truly cancelled once the user has closed the screen or press the physical end on the device?
    1. ScreenUiEngineAttachedListener
    2. onClose
    3. keyDown
    4. .. .or all this?

  Is it advisable to use?

  Thank you.

  Do not start and cancel the timer - create, plan and cancel TimerTasks!

  Have a static _t private timer = new Timer(); -This way, you will have just one per application (at least for this class), statically initialized. Cancel the timer just before you shut down the application. Plan all your TimerTasks on a single clock.

  Your question then becomes "when can I plan and cancel my TimerTasks?" It depends on your specific needs for your TimerTasks, but the right places would be:

  Annex:

  Screen.onDisplay () / onUiEngineAttached (true)

  Screen.onExposed () (if you go the road of onObscured() below)

  Screen.keyChar () / keyDown() / keyControl() (if you go with the DeviceInfo.getIdleTime check () below)

  Screen.navigationClick () (see above)

  Screen.touchEvent (see above for touchscreen devices only)

  To cancel:

  Screen.onUndisplay () / onUiEngineAttached (false)

  Screen.onObscured () (pairs with onExposed() above)

  If (DeviceInfo.getIdleTime () > yourTimeoutValue inside) TimerTask.run () method - must also deal with the above-mentioned keyChar/keyDown, the navigationClick() and the touchEvent() then.

  One thing to keep in mind - a TimerTask actively running will drain your battery pretty quickly by preventing your phone to enter a low power mode, so don't forget to cancel it when you don't need. On the other hand, a timer 'active' without any scheduled on that, go to sleep just forever, so don't bother killing/recreate.

• Size of the application

  Hello

  This question is not the actual size, but the size of an application file.

  From past experiences, I noticed that BlackBerry apps are usually around 500 KB to 1 MB in size. As a matter of design should we limit to file size. I know its always good practice to do so, but given that the PlayBook will be much more memory, it is ok for an application to spread 2 - 3 or even 5 MB range?

  Thank you!

  GP

  Hey gpatton,

  its seems perfectly fine for an application to be of this size. in fact I would not be surprised if they grow (e.g., games and graphics). to be more precise in the webcasts its been said that the apps will have a size limit of 64 MB. 3-5 MB is therefore well below that. hope that helps. Good luck!

• The synchronized application: creation of menu items for each call to makeMenu SQLite?

  A matter of design, which is a good implementation of the following?

  My requirements are:

  1. Have a different menu for each screen.
  2. The battery of the screen will be high both a single screen.
  3. To switch screens, the user must use the menu.

  I created the menu for each screen by the following:

  protected void makeMenu(Menu menu, int instance) {
  super.makeMenu(menu, instance);
  Database db = null;
  try {
  
    db = ResourceManager.getConnection(true);
    mCoffeeDao = new FlavshellDao();
    mTabDao = new FlavmDao();
  
    if (instance == Menu.INSTANCE_CONTEXT) {
        Bitmap logoWbg = Bitmap.getBitmapResource(resource
                .getString(ScoreCenterResource.logo));
        Image image = ImageFactory.createImage(logoWbg);
  
        gMenu.setIcon(image);
        nMenu.setIcon(image);
        aMenu.setIcon(image);
        shMenu.setIcon(image);
        avMenu.setIcon(image);
        logoMenu.setIcon(image);
  
        menu.add(gMenu);
        menu.add(nMenu);
        menu.add(aMenu);
        menu.add(shMenu);
        menu.add(avMenu);
        menu.add(logoMenu);
    }
  
    if (instance == Menu.INSTANCE_DEFAULT) {
        menu.add(logMenu);
        menu.add(exitMenu);
  
        shell = mCoffeeDao.findAll();
  
        for (int i = 0; i < shell.length; i++) {
            Flavshell m = shell[i];
            /**
             * Ordering parameter, lower values are placed closer to the
             * top of the menu screen.
             */
            item = new FlavMenuItem(m.getPk(), m.getFlavortitle(),
                    200 + i);
            menu.add(item);
        }
  
        String[] params = {
            Integer.toString(this.mCoffeePk)
        };
        Flavm[] m = mTabDao.findByDynamicWhere("FLAVOR = ?",
                params);
  
        if (m.length > 0) {
            for (int j = 0; j < m.length; j++) {
                if (m[j] != null) {
                    Flavm specificMenu = m[j];
                    item = new FlavMenuItem(specificMenu.getPk(),
                            specificMenu.getTitle(), 100 + j);
                    menu.add(item);
                }
            }
  
            menu.add(MenuItem.separator(105));
        }
    }
  
  } catch (FlavshellDaoException e) {
    Logger.logEventError(e.getMessage());
  
  } catch (FlavmDaoException e) {
    Logger.logEventError(e.getMessage());
  
  } finally {
    try {
        db.close();
    } catch (DatabaseException e) {
        Logger.logEventError("Menu Database error: " + e);
    }
  }
  

  I created a database with a single link object read in the following text:

  public synchronized static final Database getConnection(
      final boolean readOnly) {
  Database db = null;
  try {
      URI userFileURL = URI.create(System
              .getProperty("fileconn.dir.memorycard")
              + resource.getString(CoffeeHouseResource.db_path)
              + resource.getString(CoffeeHouseResource.db_name));
  
      db = DatabaseFactory.open(userFileURL, readOnly);
  
  } catch (MalformedURIException e) {
      Logger.logError("Get connection: URI: " + e.getMessage());
  } catch (ControlledAccessException e) {
      Logger.logError("Get connection: Controlled Access: "
              + e.getMessage());
  } catch (DatabasePathException e) {
      Logger.logError("Get connection: Database Path: "
              + e.getMessage());
  } catch (DatabaseIOException e) {
      Logger.logError("Get connection: Database IO: "
              + e.getMessage());
  } catch (Exception e) {
      Logger.logError(e.getMessage());
  }
  return db;
  }
  

  My question is,

  • If the user to call the menu as happening in any other activity database (SQLite), making the synchronized method will prevent database locks?
  • Should I design my makeMenu got a call from the database at all? It seems to me that all this works, this could potentially cause a lot of traffic on the SQLite database?
  • I thought to move the menu to a vector and storing the menu in storage Runtime, but I read that the Runtime Storage has the memory leak issues?

  Your basic design principles seem well - I have a number of applications that actually have "tabs" screens, so clicking on a tab or using the menu swaps from one screen to the other.  You can just push the new screen and pop, that. In fact, here's a menu from a screen, that I am currently working:

  _StatsMenu MenuItem = new MenuItem ("Stats", 110, 10) {}
  public void run() {}
  UiApplication.getUiApplication () .pushScreen (new StatsScreen (System.currentTimeMillis (), 0));
  UiApplication.getUiApplication () .popScreen (BaseScreenWithToolBar.this);
  }
  };

  While I haven't looked in detail, but your code dangerous look.

  I've never tested, but I suspect makeMenu is called using the EventThread.  Doing things like database connections in this method will cause you problems.  Do not.  If you need to query the database, do so before makeMenu is called.  Starts a separate Thread and set the appropriate menus if this Thread is not finished.  or block the user interface, until he made.

  In addition, I wouldn't do it in makeMenu:

  Bitmap logoWbg = Bitmap.getBitmapResource (resource
  .getString (ScoreCenterResource.logo));
  Image image = ImageFactory.createImage (logoWbg);

  You can do it once to the creation of the screen, or better yet, once at startup and save it in a shared domain.

  If you do not want to create a dynamic menu, then use the normal persistent store - or check out the details for the database once at startup and use them after that RAM.

  And avoid lifting heavy objects in makeMenu.

  I hope this helps.

• Fabric of interconnection to Bay iSCSI connection

  I need to connect my fabric for interconnection to a new array of iSCSI.  There are a number of blades UCS that need to connect to iSCSI LUNS. The FI is currently connected to the rest of the network through switches to Nexus 7000 K.   Should I I plugged directly from the FI to the iSCSI array or go through the 7000 K and then to the table?

  Hi, this is more a matter of design, you need to think about what will need access to the storage ISCCI matrix. For example, if a single UCS blaes will have access to this storage group, you can consider plug directly such as iscsi traffic must pass your N7Ks if the two fabrics are active. If you want another type of server such as HP or IBM to access storage, you should consider to connect the storage array to the N7Ks if your tissues are configured in end-host mode. Again, this will depend on your current implementation

• VI on a BladeCenter S for remote offices

  Hi everyone, it's a matter of design because I don't know much about BladCenter S.

  After reading this article, I become curious of this technology (http://it20.info/blogs/main/archive/2008/11/14/162.aspx) because it really can solve some problems, especially in remote offices.

  My questions are:

  (1) is the redundant eth switch? Article that I linked above, it seems that it is but to read the IBM documentation looks that all EPF blades are related to bay1 only, so only one switch eth (which would make not much sense);

  (2) is the internal memory, support for vmware? Could not find information on that, nor what hw config is required to share LUNS with blades;

  (3) is it possible to have 4 network adapters and interfaces 2 sas (from FC interfaces for it matters) without developing the blade server? I can't expand the blade increases the number of cards you can add but then each blade will take twice as much to place in the chassis, which is not good in a small bladecenter as the S

  I suppose to use HS21 XM blades just like I would on another chassis. Ibm documentation I found here (http://www.redbooks.ibm.com/abstracts/sg247523.html) and here (http://www.redbooks.ibm.com/abstracts/sg247682.html?Open) was very detailed, but confused about what I mentioned in my questions.

  Yes, ETH that a switch can be used, but then you get only half of the network connections.  Each blade is divided between the i/o modules, so if you have expansion for the blades ethernet card, and only have 1 switch, and then you get only half of the available network adapters.

  The IBM SAS HBA card is in the HCL: http://www.vmware.com/resources/compatibility/search.php?action=search&deviceCategory=io&productId=1&keyBasic=ibm+sas+&maxDisplayRows=50&key=ibm+sas+expansion&release%5B%5D=-1&datePosted=-1&vid=&did=&svid=&ssid=

  And Yes, I believe the extension ethernet 2 ports and the sas of maps extension and in a format that can fit into the PCI - X slots without the blade who need to take two slots, according to the guide.  I'm sure your IBM representative can go over this with you as well.

  -KjB

• List of polling for changes in State?

  Looking for a little advice on how to proceed in what is probably a matter of design until I commit 10 000 strikes in the wrong direction...
  
  I have a LinkedList. Objects stored in the list include a calendar control indicating if the object was created. Objects may expire while remaining in the list and should be deleted, as they expire, vivid sense is important, more or less. I came across a ConcurrentModificationException when I tried to use an iterator, which prompted me to change the underlying data structure of a CopyOnWriteArray, but unfortunately, it doesn't support moves.
  
  What I have have written works, but looks like it's a problem ticking. My intention is that only one thread is the queue at all, and LinkedList allows you to perhaps 30 or 40 items.
  
  Issues related to the:
  
  (1) is there a better way to achieve what I want than what I wrote? The answer is definitely 'Yes', but I am at a loss about where to go from here where I am.
  
  (2) when an object expires, I would like to notify the main process of this thread. I think to implement Observerable since it seems to involve less code that throw events and more orderly and structurally easier to follow than a simple reminder.
  
  Thanks in advance for any ideas you can offer. Code sample follows.
  

      public void run()
      {   
          while(flag)
          {
              for(int count = 0; count < queue.size(); count++)
              {
                  calendar = Calendar.getInstance();
                  Widget w = (Widget)queue.get(count);
                  Calendar wCal = w.getCreateTime();
                  
                  if((calendar.getTimeInMillis()) - (wCal.getTimeInMillis()) >= w.getWaitTime())
                  {
                      // Timer expired
                      System.out.println("PollingQueue: Widget " + w.getName() +
                              " expired.");
                      w.setState("EXPIRED");
  System.out.println("Removed " + w.getName());
                      queue.remove(count);
                      
                  }
                  else
                  {
                      if(calendar.compareTo(wCal) > (w.getWaitTime() * 0.8))
                      {
                          System.out.println("PollingQueue: Widget " + w.getName() +
                              "urgent.");
                          w.setState("URGENT");
                      }
                  }
              }
              try
              {
                  this.sleep(getPollingFrequency());
                  System.out.println("PollingQueue: Thread sleeping");
              }
              catch(InterruptedException e)
              {
                  // Thread interrupted
              }
          }

  873798 wrote:
  What I have to synchronize LinkedList itself?

  This question is meaningless, because you cannot synchronize an object. What I said earlier, is that you must synchronize all access to this list. In other words, if it must be accessed by multiple threads, every bit of code that touches it must be synchronized on the same lock. In general, this lock will be the list itself.

  For example:

  synchronized (list) {
    list.add(newItem);
  }
  
  ...
  
  synchronized (list) {
    for (Iterator it = list.iterator(); it.hasNext(); ) {
      Thing thing = it.next();
      if (thing is old) {
        it.remove();
      }
    }
  }
  

  Note that you can save yourself some of that by calling myList list = Collections.synchronizedList (originalList) immediately after the creation of the original list and using myList everywhere. You have to sync around the simple operations like add(), but you will still have to sync around the iteration.

  Because only the main class and mark this thread can visit to LinkedList, synchronization is not a big problem in this case, but this solution would not sorely lacking where multiple threads try to access LinkedList?

  So you are assuming that the timing "makes things slowly." All that can be the case when it is poorly managed, is not generally Yes. And, more specifically, if you need to sync, you need. You can't jump just because you think it's "too slow." If your code is the wrong thing, it won't matter what happens quickly.

  I recommend you read Goetz or Lea books Java concurrency. There are many things that you just guess and make assumptions about what you absolutely need to know if you are going to write multithreaded code and that it is too difficult to explain in a forum.

• How to hide scroll bars in Muse

  Hi all

  I was wondering y at - it an option to hide the scroll bars in a Web site built with Adobe Muse? I know that Safari hides the default scroll bars but no older browser. I just want to hide for a matter of design.
  Thanks to you all!

  Find out what...

  Disabler Widget

  It's free, click on the free download for the file.

• 11 GR 2 Oracle home path

  Hello
  
  It is a matter of design rather than a technical problem.
  
  With Oracle OFA paper [http://download.oracle.com/docs/cd/E11882_01/install.112/e17162/appendix_ofa.htm#BABHAIIJ].
  
  The grid and the Oracle Home are suggested to be:
  /*/app/Oracle/product/11.2.0/DB* for Oracle Database 11g oracle directories
  /*/app/Oracle/product/11.2.0/grid* oracle home directory for the Infrastructure of grid of Oracle 11 g for a stand-alone server for the oracle user
  
  
  However, I have read that there 11.2.0.2 group of patches, which is a new full version.
  Oracle recommends to have the upgrade of Out-of-place (this is Oracle recommended to apply a patch to the game. You install the hotfix in a new, separate Oracle home location) and this is Oracle approach in the future.
  Grid, it must be upgraded out of place.
  
  So does make sense that I should install the Oracle home /*/app/oracle/product/11.2.0.x/grid* and '/*/app/oracle/product/11.2.0.x/db* '?
  
  
  Thank you.
  
  Rocky

  Welcome to the forums!

  This is a valid and good question. I think that your approach is logical, given that future update sets (11.2.0.3 etc.) would be also (I hope) released that installs complete and will very likely need to be installed in separate houses.

  HTH
  Srini