Personally, if I were starting something new today, I'd lean towards Python over REXX in most cases.
That's not because REXX is a bad language. In fact, REXX has served the mainframe world extremely well for decades and is still very effective for many z/OS scripting and automation tasks.
The advantage Python brings is the broader ecosystem around it. You get access to a huge developer community, countless libraries, API integration, automation frameworks, data analytics, and increasingly AI-related capabilities. It's also a skill that transfers across platforms, whether you're working on IBM Z, Linux, cloud, or distributed environments.
That said, I wouldn't advocate replacing working REXX code just for the sake of modernization. If an existing REXX solution is stable, maintainable, and doing the job, there's often little business value in rewriting it.
My view is simple: keep REXX where it makes sense, but for new development and automation projects, Python would generally be my first choice because it gives you more flexibility and future-proofs your investment in skills and tooling.
--------------------------------------
Abhishek Anand
IBM Z & LinuxONE Brand Leader,
SBM, Riyadh,
Saudi Arabia
--------------------------------------
------------------------------
Abhishek Anand
------------------------------