Yeah, me too. I reluctantly backed into postfix because I wanted the order a computer computes a program in to track the order a human reads a program in. That would help show intermediate results and reduce the need for the human to "play computer". Neither infix nor prefix have this property, and I wanted to stick with a single dimension so that programs continued to stay dense.
I also was trying to come up with a clean HLL that can also double as a shell for the Mu computer, so people don't have to learn more than 3 languages to browse all the code for the computer. Postfix feels like a good fit for shell languages since you can add operations to a command as you think of them.
The big con was the lack of variable names in postfix functions. I feel like I've resolved that fairly satisfactorily.
Thank you!