John Cowan wrote: > Paul Jackson wrote: > The less ed(1) changes, the happier I am.
+1. People are not using ed because it has color syntax highlighting, integrated debugger, or vcs integration. People use ed because it is The Standard Editor that one can always count on being there and operating exactly in a standard way. > I sympathize entirely with that point of view. However, if all non-Posix > features are gated behind command-line options, particularly GNU-style > options like --pcre, then I think everyone's interests are served: > > - If you want PCRE matching, use alias ed='ed --pcre'; > - If you don't, don't use an alias; > - To bypass the alias temporarily for performance or whatever reasons, > type \ed to the shell. Not that long ago, well it was pre-pandemic times when we used to meet-up in person, I used ed to rescue a friend's Ubuntu laptop that he had messed up while at the meet-up! He had no functional editor that we could find other than ed. I don't know why but he had the packaged ed installed. I am guessing some DE dependency pulled in a dependency and on down the line and ed was installed. I was able to use ed to rescue his system! No amount of local custom building could have worked in that situation. Expanding ed to include creeping features is an insidious path. Where does it stop? If Emacs or Vim are the obvious goals then why not simply use Emacs/Vim now? Why try to mutate ed to be down that path instead of simply using the editor that supports these features in the first place? And also if so then why start with ed instead of editors like nano that are already further along the path to that goal? > It's unlikely you'd type such an extension by mistake, unlike (say) ed -P. > > I already use alias ed=ed -p '* ' because I like prompts, but if I'm > running an ed script, I use \ed <foo.ed. This is not really the same thing. Because these are features that are portably available in every version of ed. Bob