Hi y’all, > I want to write portable ed(1) scripts that work on both GNU and BSD > ed(1)-s. The only hurdle to that so far is word boundary regex, which is > \< / \> on GNU and [[:<:]] / [[:>:]] on BSD. Both are extensions beyond > POSIX, but ultimately useful. I’m using these. And, to make my ed(1) > scripts portable across implementations, I have to preprocess them with > yet another script (so meta) to replace GNU-style word boundaries with > BSD-styles ones. Something like: > > H > !# Meta boundaries for ed-generated scripts > g|\\\\[>]|s||\[\[:>:\]\]|g > g|\\\\[<]|s||\[\[:<:]\]|g > !# Regular boundaries > g|\\[>]|s||[[:>:]]|g > g|\\[<]|s||[[:<:]]|g > w > q > > Is there maybe a possibility that GNU ed(1) (as the one most expansive > and portable) might support BSD-style word boundaries? That’d make a lot > of scripts immediately portable across major implementations, bringing > much joy to the people.
Any response? Antonio? (CC) Best of luck, -- Artyom Bologov https://aartaka.me
