Thank you for the great article.

I'm interested in getting back the fixed input. I know that nimbleparse tool outputs an AST and error recovery sequences. Is there a way to make the tool apply a recovery sequence to the input string and return the new valid string automatically?

The Cornell PL/1 compiler (actually a pedagogical variant they called PL/C) for the IBM 360 attempted to repair syntax errors so your compilation could continue as long as possible. It would insert tokens to try to reboot the parse, with occasional success.

This was in the 1970s with punch cards, the bad old days, when computer time was more precious and you literally had to stand around the printer waiting for the printout while your job went through the queue. It was a great idea.

@Rob

The Cornell PL/1 compiler (actually a pedagogical variant they called PL/C) for the IBM 360 attempted to repair syntax errors so your compilation could continue as long as possible. It would insert tokens to try to reboot the parse, with occasional success.

Interesting! Do you know if PL/C was the first to do this? From memory (please apply all the normal caveats) I tried, but didn't succeed at, working out when/what the first error recovery approach was. Indeed, I couldn't even work out where "panic mode" first originated, which seemed like it should have left a clear paper trail!

The older I get, the more I realise how much of that mid-ish period computing history is in people's heads and how in danger it is of being lost...