One of the fun things about EBCDIC is that 370 assembler has opcode-level support for converting an EBCDIC-encoded numeric string into an integer (and maybe the

One of the fun things about EBCDIC is that 370 assembler has opcode-level support for converting an EBCDIC-encoded numeric string into an integer (and maybe the other way around too, it's been a while). This is one of two things I remember about my now-ancient 370 assembler knowledge. The other is that there is no built-in support for maintaining a call stack. It is up to each subroutine to handle this and there were some weird declarations around this to indicate whether a subroutine was reentrant, the definition of which escapes me now. And people shouldn't criticize EBCDIC too much, after all Windows still dumps a lot of crap in legacy 8-bit coding that can cause applications to break (there was a recent post on HN about someone being unable to run the IntelliJ debugger because of an accent in their username). At least EBCDIC is clear about its limitations.¹ ⸻⸻⸻ 1. I'd be remiss if I didn't point out one other EBCDIC weirdness: It has two vertical bars, | and ¦ which always caused complications in translations between EBCDIC and ASCII. IIRC, ¦ was the more common symbol in EBCDIC coding but some converters wanted to translate | to | instead (or maybe it was the other way around—the last time I did IBM big metal was 30 years ago).