IBM Mainframe Forum

[RODders]

I wonder if anybody has migrated from IBM PL/I for MVS & VM to enterprise Pl/1 and faced any issues running mixed modes of executables

[Akatsukami]

Yes; a former client converted in 2006-2009, IIRC. My recommendation is that no run unit be of mixed MVS/Enterprise routines. You do know what your run units are, yes?

(my client didn't know what its run units were; gathering that information was one reason that the conversion took so long.)

[RODders]

The issue is we have a massive amount of common pl1 code used by various business areas thus they wanted us to run with mixed modes

[Akatsukami]

yes, that was the case at my client, also. That was why we had to establish what run units we had; we had a shipload of CRUD routines, often at the end of a chain of dynamic calls (which MVS PL/I would not support, so we had a HLASM module doing the intermediate calls), used by many different load modules. The load modules and all intermediate modules had to have all been converted, or else we got S0C4 abends. We ended up cloning the source, converting the clones, testing them, and then issuing the cloned code to production.

Please tell your business partners that a very senior developer who has been through exactly this conversion says running mixed run units will not work. Period. Full stop.

[RODders]

Thank you very much this

[prino]

I wonder if anybody has migrated from IBM PL/I for MVS & VM to enterprise Pl/1 and faced any issues running mixed modes of executables

Big bang is, in essence, your only "safe" option. At one client we changed to compile procedures to recompile changed source with both compilers (old version still went into production), but any RC>=8 on the new compiler, and believe me, running with very restrictive compiler options, we had zillions, would force developers to look at the code, make a note of the problems and obviously correct them - with somewhat less restrictive compiler options we would have had fewer problems, but we wanted to use the exercise to also give us a clean code-base.

Decode my email address, you've probably heard about gmail , for the EPLI 4.3 options now in use.