Hacker News Reader

Comment by luigi | Hacker News Reader

Here's the Twitter discussion alluded to in the post:

https://twitter.com/harthvader/status/293829635823792128

Main justification is the nicer syntax. I agree it's nicer, but would approach it by wrapping sed, not re-implementing in Node.js from scratch.

I don't think that excuses the ridicule, though.

View on HN

The maximist point of view is to wrap sed, yes. But what also came to my mind was what if he/she just wanted to only learn, I can think of no better way than to try to write it 'yourself' in the language of your choice, if only as a learning exercise.

All to often we in technology assume the maximist point of view, possibly completely discounting another dynamic point of view or some inspirational reasoning (of the originator).

*edit: spelling

Me neither. It's not as if klabnik was asked to use it. Plain old rude, in the name of defending no principle in particular.

How would you get sed to output the files and lines on which replacements happened?

How about using diff and sed -i?

  # This version is obviously unsafe
  verbose_sed() {
    sed -i.old -e "$1" "$2"
    diff -u "$2.old" "$2"
    rm "$2.old"
  }
  
  verbose_sed s/vim/emacs/g rant.txt

You could then use all the tools that we already have for working with patches: colordiff to colorize the output, diffstat for a summary of changes, patch -R for reverting the changes, and so on.

Of course, you're more likely to than not already using version control, which gives you all this and much more, even if you were to use vanilla find+sed.

I don't think rewriting as a wrapper around sed would be an improvement. That would put you in the business of translating one regexp syntax into another which is probably more error prone.

Reply on news.ycombinator.com