Sometimes I find something that I think is cool, it may be something small, like this, but I feel it is worth sharing.
I was asked about how to find the minimum value from multiple columns. SQL offers us the MIN function for returning the smallest value from a column. I can use that to get the smallest value from multiple columns, and the smallest value from all of those columns.
I have a file with three fields containing the following numbers:
Sometimes it is just "better" to use a list API than it is to use the SQL View that "replaced" it. This week has given me an example of this. I was asked to create a program that would allow users to email themselves reports. I knew I would have to present the users with a list of spool files from which they could select the one they would want to email.
"No problem", I thought, "I will just use the Output Queue Entries SQL View."
I created a View over the OUTPUT_QUEUE_ENTIRES View, and then created a program using a multiple row fetch to get more than one row from the View. In the "test" IBM i partition the program ran a bit slow the first time I ran it, then ran faster each time after as the DB2 query engine created a temporary access path for me. Being a "test" partition there are not many output queues or spool files when compared to the "live" partition. I moved the objects I created to "live" and called the program there. It took more than 20 minutes to load the subfile with the list of user's spool files. With several hundred users and a very large number of spool files (many thousands, does anyone ever delete spool files?) in the "live" partition it was searching through all of them to find the few spool files belonging to me. The time taken made this method unacceptable. Back to the "drawing board".
The web page that was originally linked to no longer exists, therefore, rendering this page obsolete.
One of my colleagues was reading the post about initializing variables when defining them and asked a question that I had not considered when writing that post.
"What if I change the value in the variable," he asked, "can I change it back to its original value?"
Fortunately there is in RPG an operation code that will do this. The RESET operation resets the value of any variable or record format back to the value it was at the initialization phase of the program. It is not new operation, I have used it for as long as I have programmed on the AS400 - IBM i.
I have Steve Pitcher to thank for sharing this image so that I could post it here. It is from IBM's KnowledgeCenter showing the different types of saves that can be performed using the various save commands, on the right side, and the options on the SAVE menu, on the left.
Included in the latest Technical Refreshes for IBM i 7.3 and 7.2 is a SQL View and Procedure to help us monitor the licenses of the software on our IBM i.
IBM uses Software License Keys with certain IBM i products to ensure that I have paid the appropriate license fees to use these products on an IBM i. I know of several third party IBM i software vendors who also use this method to stop the unauthorized copying of their software to unlicensed servers. It is all standard stuff to prevent the distribution of unauthorized software.
I frequently get messages asking why does IBM call it IBM i when it is just an AS/400, or even System/38, by another name. Talking to other authors and IBM-ers I find that I am not alone to receive communications like this.
Steve Will, chief architect for IBM i, has felt compelled to dispel this fallacy. Earlier this month he made an online presentation, sponsored by HelpSystems, explaining how IBM i differs and is a lot more than AS/400. The video can be watched on the sponsor's website here.
The video lasts just over an hour, and, in my opinion, is worth watching to learn the evolution of the operating system to become the thoroughly modern IBM i.
The germ for this post's subject came from a two part question I was asked:
- How can I know which field the cursor is in when Enter is pressed?
- How can I position a cursor on some field without using indicators? No indicators are available in that very old screen.
Fortunately there are ways to do this by adding a few keywords and fields to the display file's DDS. In these examples I am going to use one display file with three screens, each showing a way to do what has been asked, and the RPG code needed too (and I am going to cheat on the second part).
Last week at OCEAN User Group's Technical Conference I attended a couple of sessions by Scott Forstie, IBM's Business Architect for DB2 for i. I noticed in his presentation he had used "Db2" rather than "DB2".
When I asked him about this he explained that last week (Monday July 17, 2017) the DB2 family had been officially renamed "Db2". The Db2 family is:
I found that that Db2 page on IBM's website has already been changed, see here.
Now I need to remember to use the new name when writing about it in this blog.
I needed to determine, first, if a certain character was present in a file's field, and then how many times it occurs. Using RPG's file input I would not be able to do this without reading and checking every record's field. Could this be done easier in SQL? Yes.
I am going to give examples of various ways of being able to tell if a certain character is in a field in a file, using increasing complex methods I found. At first I will just give the SQL Select statement for each, and at the end the SQLRPGLE code I created to do this in a program.
I have been asked to write some more about using SQL embedded in RPG programs. One question I have been asked by several people is how to update a row from a file that had been previously Fetched, just like READ and UPDATE using RPG's native database access. I thought it would be useful to show examples using both RPG native database access and SQL.
First I need a file, as this is for example purposes it very simple, just two fields and one key field:
Nulls are something we are all going to have to get use to handling as we move from DDS files to SQL tables, or start using data from DB2 of i or other non-IBM i databases. Recently I was working on a project with a SQL Table with more than 300 columns. If there was no data in a particular field it was null. If you think about it null is a better alternative than blank or zero as those values could be valid values for the contents of that column.
I have written about handling null in SQL in an earlier post, but in this scenario the method I described to define the null byte map was not practical. I am not going to list 300+ subfields in a data structure array for the null indicators for the matching data structure array I would be moving the SQL Table's data into.