Replace QTEMP library with User Libraries

[Guest.Visitor]

Have your program call a CL program to create your library. The CL program will have a data area to have a control number 1 - 999999999. When the CL program runs, retrieve the data area and update it by add 1 to it. Use the retrieved number and put a letter is the first position and that is the name of the new library. You can pass the number (or library name) back to the web program and use it later to delete the library just before the web ends. When the number reaches the all "9", make sure the old libraries are deleted. You can use a 4 digit date with seq of 5 "9" to id the library easier. You can have the CL program get the 4 digit date and when the date changes, update the seq # back to 00001. You can use this CL program to build libraries for many apps by passing the first character and setting up multiple data areas (for seq #) per app. If you do not have more than 9999 files for a day, you can use a two letters instead of 1 for the library name. If you can not control when the app finishes, you can have another CL program control deleting the libraries based on a date routine.

[NeilD]

Replace QTEMP library with User Libraries

My (interactive green screen) application depends heavily upon QTEMP to hold work files and DTAARA's for each session. I need to replace QTEMP when we develop the system as a web application. Does anyone have any ideas? Rather than change every program to replace dtaara's with LDA or a user file, it occurred to me that I could create a user library for each user session. This would then need deleting as each session ended. Anyone done this? Thanks for any help you may have.

[cfuture1]

Replace QTEMP library with User Libraries

You could write a simple enough program to read and write the data to an IFS. First decide what common identifying values will make the data unique to the work flow. QTEMP served that purpose for these jobs. This might be (1)a user name, (2)user name and date and time of initiation, or (3) user plus date plus a sequence number within the date. These values can serve as directories in the IFS. To distinguish their purpose from other IFS objects, you could keep them together within a parent directory, for example, '/XTEMP'. So when a session initiates, you would have "/XTEMP" already set up, and so create a directory for the user if not already there: "/XTEMP/BOBUSER". Then create a directory for this "Web session" for the occasion, say "/XTEMP/BOBUSER/J20071115-090909", for example, incorporating the date and exact time. I like the idea of this naming convention because it may make maintenance a little easier, by telling at a glance how old it is. Check the directories to see if they're there before creating them. For initiation and your "lowest-level" directory to be session-specific, you check to see if it's there and if it's there, then you get a new time for the name and re-try. Or add sequence numbers. At that point you can just use the data-area name for the new IFS file you create for the data-area data. No matter how you do it, the resulting libraries and/or IFS directories-and-files will require maintenance checks to make sure your sessions cleaned up after themselves when they ended. --Alan