Installing Unbreakable Enterprise Kernel 4 on Oracle Linux 7

As announced here the Unbreakable Enterprise Kernel 4 is now available for Oracle Linux 6 & 7.
I was actually building a baseline APEX VM based on my personal preferences late last year.
To safe others and me the searching here is a quick writeup of the steps involved.

Start by making UEK4 repository for Oracle Linux 7 known to yum.
Use your favorite editor to edit the file /etc/yum.repos.d/public-yum-ol7.repo.
Then either add below lines to the file or replace the ones for ol7_UEKR3 with them.
Actually you cn copy the repository ol7_UEKR3 and just replace every 3 with a 4.

[ol7_UEKR4]
name=Latest Unbreakable Enterprise Kernel Release 4 for Oracle Linux $releasever ($basearch)
baseurl=http://public-yum.oracle.com/repo/OracleLinux/OL7/UEKR4/$basearch/
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-oracle
gpgcheck=1
enabled=1

Now run yum update and verify that you actually get the version 4.
Depending on your internet connection's bandwidth this is a good time to grab a coffee.
You should at least see below line, otherwise you won't get the new kernel.

kernel-uek                      x86_64             4.1.12-32.el7uek                    ol7_UEKR4               43 M

Review the output from yum and then confirm with y.
Wait until yum has downloaded and installed all pending packages.
After everything has been installed reboot to activated the new kernel.

If you're running the system within VirtualBox like I do remember to reinstall the VirtualBox Guest Additions. You need to do this every time you boot into another kernel because they are compiled against the currently active kernel.
In the VirtualBox VM window go to Devices->Insert Guest Additions CD image.
Then mount the cdrom device if it isn't mounted automatically, then reinstall the additions.

[root@ol7db12c ~]# mount /dev/cdrom /media/cdrom
mount: /dev/sr0 is write-protected, mounting read-only
[root@ol7db12c ~]# /media/cdrom/VBoxLinuxAdditions.run
Verifying archive integrity... All good.
Uncompressing VirtualBox 5.0.12 Guest Additions for Linux............
VirtualBox Guest Additions installer
Removing installed version 5.0.12 of VirtualBox Guest Additions...
Removing existing VirtualBox non-DKMS kernel modules[  OK  ]
Copying additional installer modules ...
Installing additional modules ...
Removing existing VirtualBox non-DKMS kernel modules[  OK  ]
Building the VirtualBox Guest Additions kernel modules
Building the main Guest Additions module[  OK  ]
Building the shared folder support module[  OK  ]
Building the OpenGL support module[  OK  ]
Doing non-kernel setup of the Guest Additions[  OK  ]
Starting the VirtualBox Guest Additions [  OK  ]
Installing the Window System drivers
Could not find the X.Org or XFree86 Window System, skipping.

Have fun with the new kernel!

Coloring Cells in an Interactive Report

This is a quick post on something that came up on slack this week.
The task was basically to highlight certain cells in an interactive report based on the values in another column. While the highlighting feature in the interactive report is great that's one thing you can't do with it. You have the choice of highlighting the whole row or the cell in the column on which you based the condition. Another situation where the builtin highlighting isn't enough is when you need some complex conditions involving more than one column.

Took some iterations, but here's one solution to this.
I'll first outline the building blocks and then explain how they work together including some failed attempts on the way getting to this.
Interactive Report Query:

SELECT empno,
       ename,
       job,
       mgr,
       hiredate,
       sal,
       comm,
       deptno,
       CASE WHEN sal <= 1500 THEN 'data-style="background-color:green"' 
            WHEN sal > 1500 AND sal <= 3000 THEN 'data-style="background-color:orange"'
            WHEN sal > 3000 THEN 'data-style="background-color:red"'
            ELSE NULL
       END css_style
  FROM emp

Then set the css_style column to hidden and set the "HTML Expression" attribute on the column you want colored to something like the following.
Note: I use the SAL column here, but you can really use any column.
HTML Expression of column SAL:

<span #CSS_STYLE#>#SAL#</span>

The last bit is some JavaScript code which you put into a dynamic action firing on event "After Refresh" for the interactive report region. Make sure to also set "Fire On Page Load" to yes, otherwise you have no coloring after page load.
JavaScript Code:

apex.jQuery("span[data-style]").each(
  function()
  { 
    apex.jQuery(this).
      parent().attr( 'style'
                   , apex.jQuery(this).attr('data-style')
                   ); 
  }
);

If you don't want to read the full story you can stop reading here and have a look at the example app to be found at https://apex.oracle.com/pls/apex/f?p=83537.

Still reading?
Let's take a look at how we got to a working solution and how it does it's magic.

In the query we are laying the foundation by adding a column which computes the needed style for us. You might wonder why there is "data-style" instead of simply "style". Let me show you how the report would look if you would just be applying a style to the span element.

Just styling the span element

This doesn't look good and the reason is that the span element is an inline element only using the space it needs and not filling all available space as block elements do. Next try would be setting the span element to behave like a block element by adding display: block to the style definition.

Styling the span element with added display:block

A bit better, but this still isn't coloring the whole cell. To keep it short there where some other tries using some CSS-Wizardry which didn't work out too good either.
Back to the "data-style" thing. You are allowed to add your own custom attributes to HTML elements if you prefix them with "data-". Browsers will ignore them but from JavaScript or in this case jQuery you can access and utilize those attributes.
Let's have a closer look at the JavaScript code and walk through the logic.

1. Select all span elements which have an attribute called data-style

   apex.jQuery("span[data-style]")

2. Loop through the jQuery array and call an anonymous function for each item
   

   .each( function() {

3. Find the current item's parent element (that's the for the cell)
   

   apex.jQuery(this).parent()

4. Set the parent's style attribute using the span's data-style attribute
   

   .attr('style', apex.jQuery(this).attr('data-style'))

Finally we had something which looks very much like the inbuilt highlighting:

So long folks, this post actually got longer than I expected.

Happy IR-Hacking :-)

Using CodeMirror in your APEX Application

I just had a look at a blog post by Connor McDonald where he's asking for opinions how code should be presented on the AskTom website.
In my opinion I would love to have the line numbers removed from the actually code and instead let the get generated by a code highlighter.
With APEX 5 the development team uses CodeMirror for the App Builder, meaning the needed library is already included with every installation of APEX 5.

Getting your code highlighted

First of all I'm assuming you put your code into some sort of a tag, otherwise it wouldn't possible to distinguish code from other text. I did a quick check on the AskTom website and they seem to use the <pre> tag, so I'll use the same to demonstrate. Let's prepare a single page for displaying nice highlighted code.

Update page JavaScript and CSS attributes

Add following to JavaScript File URLs:

#IMAGE_PREFIX#libraries/codemirror/4.4/codemirror-custom.min.js
#IMAGE_PREFIX#libraries/codemirror/4.4/mode/sql/sql.min.js

Add this to the "Execute when Page Loads" sections:

apex.jQuery(".code").each(function() 
{ 
  var myThis = apex.jQuery(this);
  var myCode = myThis.html();
  var myMIME = myThis.attr("data-code-mode");
  myThis.empty();
  CodeMirror(
    this,
    {
      value: myCode,
      mode: myMIME,
      lineNumbers: !myThis.is('.nolinenumbers'),
      readOnly: true
    }  
  );      
});

This code searches all elements having class "code" and uses the contained html as the code to highlight. The language is derived from an attribute called "data-code-mode", if you want to use something else than what is included in the sql.min.js file you need to include additional files from  the CodeMirror mode directory.

In CSS File URLs add the CodeMirror gloabl css file:

#IMAGE_PREFIX#libraries/codemirror/4.4/codemirror-custom.min.css

Finally specify this for Inline CSS:

.CodeMirror {
    height: auto!important;   
}

Adding code blocks

Now just add a new region to the page, I have used static content for now as this is just an example.
The content of the region is the following:

This is some text first...
Now show code:
<pre class="code" data-code-mode="text/x-plsql">DECLARE
  l_data VARCHAR2(32767);
BEGIN
  NULL;
END;</pre>
More text below code block.
now using a div for code:
<div class="code" data-code-mode="text/x-plsql">
SELECT user
  FROM dual;</div>
Nice, isn't it?

Finally some code without line numbers 
<span class="code nolinenumbers" data-code-mode="text/x-plsql">SELECT * FROM emp;</span>
where we just select everything from the famous emp table.

And the rendered result looks like this:

This is all more or less a quick hack, but I hope you get the idea and can make some good use of it.

Using Wallet Authentication with SQLcl

Ever since SQLcl came out I was waiting for it to support wallet authentication.
Today the waiting is over and all you need to do is grab the new SQLcl release from the SQL Developer download page. (Hint: The download link is at the bottom of the page.)
Now that you have the latest release let's prepare your client machine to use credentials stored in a wallet.

Preparing the for wallet authentication

First of all you need an Oracle client installed as wallet authentication is not possible using the thin driver. Make sure to have your client home on 12.1.0.2.x otherwise SQLcl can't make use of it.

Then update your sqlnet.ora file by adding following lines:

WALLET_LOCATION =
   (SOURCE =
     (METHOD = FILE)
     (METHOD_DATA =
       (DIRECTORY = D:\ora_wallet)
     )
   )

SQLNET.WALLET_OVERRIDE = TRUE

Make sure to update the line where the directory is set to match the location where you want to store your wallet. The last line allows the wallet to override any existing OS authentication setting.

Now modify your tnsnames.ora and add another entry which points to the database you want to connect to later. Give it a recognizable name because you will need it again when saving the credentials in the wallet. I will use "wallet_test" as the tnsname in this post.
If you use different users on a single database you have to add one tns entry for each user. The reason is that there is a one to one relationship between a credential and a tns entry.

Next step is to create the directory where your wallet should be stored.
Afterwards create the wallet by using the orapki utility from the Oracle Client:

orapki wallet create -wallet "D:\ora_wallet" -pwd "test_1234" -auto_login_local

Now store the credentials in the wallet by using the mkstore tool with following syntax:

mkstore -wrl "D:\ora_wallet" -createCredential tnsname username password

You can also emit the password and then you'll be asked for it by the tool.
Enter the wallet password and wait for mkstore to exit.

Finally test your setup by using good old SQL*Plus:

sqlplus /@wallet_test

Using the wallet with SQLcl

SQLcl is a Java application and therefore uses jdbc to connect to the database.
The latest release added support for custom jdbc connect strings and that is what we actually need.
Connecting with SQLcl using the above created credential looks like this:

sql /@jdbc:oracle:oci8:@wallet_test

The syntax might look a bit strange compared to the SQL*Plus one but you need to put the "/@" before the custom jdbc connection string and still prefix the tnsname with an "@".

Enjoy connecting to your database without specifying a password on the command line.
The main use case will most probably be for batch scripts, but I think switching to wallet authentication as much as possible also helps when you have to work in public places.

Thanks to the SQLcl team for adding another great feature.

P.S.: @Kris, the next blog post is on you. ;-)

IR Column Headings with HTML in APEX 5

Old and new world

This might be something for all of you that have used HTML in interactive report column headings .
It was a convenient way to break up long headings if you already knew the content was much smaller.
To give you an example think of a date field with format mask 'DD.MM.YYYY' and a column heading of 'Final Auditor Approval Date'.
Now compare the size of the content to the heading size, which is 10 characters to 27 characters, so you might be wasting a lot of real estate for this column.
One approach is to specify the column width and enable word wrap using css, but what if you wanted the column heading to wrap at a specific position?
Well you can play around with your css rule until you get the desired result or you simply embed an html line break into your column heading and I typically used the latter.
The minute you upgrade to APEX 5 there's a small caveat with this solution.
Let me show you an example of what I'm talking about:

Do you see the html characters in the column list when you click on the magnifying glass?
In APEX versions before 5 the tag was replaced by a blank, nowadays it's escaped.
APEX 5 has gotten way better responsive features than before, but what if even that isn't sufficient?

There's an app plugin for that

Having the immediate need to find a solution I decided to create a plugin to solve this issue.
While thinking of options one that quickly came up was using the "Single Row View Label" as I'm not using the "Single Row View" in any of my applications.
Below you see an example I used during testing:

I removed the line break from the "Heading" and then changed the "Single Row View Label" to include it. So the data is stored in the APEX Metadata and I should now be able to use it.

The plugin itself is a dynamic action plugin and puts a small JavaScript function into the page.
The function retrieves a JSON object with column ids and column headings and then iterates over the object replacing the heading. Below you find sample code for the function:

function de_itt_apexir_hdr_break()
{ apex.server.plugin
  (
    '#AJAX_IDENT#' // This gets replaced by apex_plugin.get_ajax_identifier
  , { x01: this.affectedElements[0].id } // IR id
  , { success: function( pData ) 
               { apex.jQuery.each
                 ( pData, function(index, value)
                   { apex.jQuery("#" + index + " > a").html(value); }
                 ); 
               }
    }
  );
};

The callback function runs following query to compose the JSON object:

SELECT 'C' || aapic.column_id column_id
     , aapic.form_label
  FROM apex_application_page_ir_col aapic
     , apex_application_page_regions aapr
 WHERE aapic.region_id = aapr.region_id
   AND aapr.application_id = apex_application.g_flow_id
   AND ( aapr.region_id = to_number( regexp_substr( apex_application.g_x01, '\d+') )
      OR aapr.static_id = apex_application.g_x01
       )
   AND aapic.report_label != aapic.form_label
;

So far my tests didn't show any negative impact except for the obvious extra round trip to the database. Below screenshot shows same IR as before but using the plugin.

Want to test the plugin?  Download it here and install like any other plugin.
Follow the included readme file to create a dynamic action based on the plugin.
I haven't run that many tests yet, but maybe you would like to help?
If you encounter any issues or have recommendations, either open an issue in the repository or write a comment here.

Hopefully the plugin helps you to overcome one of the small hick-ups when upgrading to APEX 5.

Using conditional compilation for APEX dependent code

I'm currently working on a new release of my PPTX Creator Package.
While adding parts which would only compile if APEX is installed in the database I thought maybe it's a bad idea to introduce a hard dependency as a lot of the code doesn't need APEX installed.
So I needed to come up with a way to allow installation without APEX, but also enable additional functions when APEX is installed.
The options I checked where:

1. Split into APEX-specific and generic package
2. Offer two versions of the package
3. Use conditional compilation

From the titel you might already know which option I chose, but let's briefly look at the implications of the options.

Option 1:

Splitting into two packages seemed like an immediate win as it feels natural for the use case. However when looking a bit further it could lead to the need of exposing internal parts of the generic package.
Running on 12c you could use the ACCESSIBLE BY clause to mitigate, but not everyone is on 12c (yet). A good blog post on the ACCESSIBLE BY clause can be found here.

Option 2:

I immediately moved over that option, just imagine the maintenance effort:
It would be a like fixing two different packages with a lot of copy-and-paste code back and forth.
No I don't want that!

Option 3:
I decided to use conditional compilation for the APEX specific features. For those of you who don't know conditional compilation yet, have a look at this Oracle Magazine Article by Steven Feuerstein from 2006 and also check out the links in that article.
In short conditional compilation allows you to exclude or include parts of the package during compilation based on conditions specified. A typical use case for conditional compilation is having one package which conditionally compiles with different code based on the Oracle Release.

So I had decided to use conditional compilation, but then I realized that I hadn't thought of how to determine if APEX is installed. Unlike the version use case, for which the database provides the DBMS_DB_VERSION package, there is no such thing for verifying installed components.
So let's start with how to generally check if APEX is installed.
If you have DBA rights you can use following query:

SELECT comp_name, version, status
  FROM dba_registry
 WHERE comp_name = 'Oracle Application Express'
;

Without DBA rights it's a bit different, as there is no ALL_REGISTRY table in the dictionary. There is a USER_REGISTRY table but that doesn't help either.
However there is something called ALL_REGISTRY_BANNERS and we can use that one with any user. To check if APEX is installed simply run the following query:

SELECT banner
  FROM all_registry_banners
 WHERE banner LIKE 'Oracle Application Express%'
;

Now let's move on to finding a means to use the query result for controlling the conditional compilation. You can also use user-defined flags for controlling the compilation process, below show how to set a flag called apex_installed for your session.

ALTER SESSION SET plsql_ccflags = 'apex_installed:1';

You can reference this flag in conditional compilation blocks within your code.
To recap, we now have a means to check whether APEX is installed and we are able to set a flag to be used by the conditional compilation.
Putting everything together I ended up with an install script similar to the code code.

SET define '^'
define HASAPX = '0'
column apx new_val HASAPX

SELECT '1' AS apx
  FROM all_registry_banners
 WHERE banner LIKE 'Oracle Application Express'
   AND ROWNUM = 1
;

ALTER SESSION SET plsql_ccflags = 'apex_installed:^HASAPX';
@@install_package.sql

Within the package you surround the code blocks with something like

$IF $$apex_installed = 1 $THEN
-- Put APEX specific code here
$END

To catch the case when the flag isn't set before compilation include an error clause like below

  $IF $$apex_installed IS NULL $THEN
    $ERROR 'Set CCFLAG apex_installed to either 0 or 1.' $END
  $END

That's all for now folks, but stay tuned for the next release of my PPTX Creator package with much enhanced functionality.

APEX IR XLSX Downloader V1.3.0

After a long time I have finally published Release 1.3 of my APEX IR XLSX Downloader.
And because I kept you waiting so long there's one goody included.

Added Features

• The Interactive Report region ID is now automatically derived if not specified or set NULL.
• Function get_version which emits the package release you are running.

Bugs Fixed

• Column header wasn't filled if user didn't give a custom name for a function column in "Group By" view.
  Now the package puts similar names to what the APEX engine generates.

New Demo Application

There is also a new demo application available for download.

You can download it here and install into your own APEX workspace.

As always the source code can be found in the GitHub Repository.

ZIP files for all my GitHub projects can now be found on my GitHub Page.

Enjoy ;-)