Paragraph command not respecting leading non-content formatting characters

[uvtc]

Given this paragraph:

> foo bar
> aoeu aoeu
> baz moo

I'd expect [p to give me:

> foo bar aoeu aoeu baz moo

but instead it gives me:

> foo bar > aoeu aoeu > baz moo

I see the same problem with //, *, #, and so on. Works fine if the leading / left-hand-side characters are only spaces.

This is with ne v3.1.2 running on a fully updated Debian "Testing".

The docs for paragraph indicate ne should support this smart Paragraph behaviour, and I thought it got this feature years ago.

I think it's an extremely useful and necessary feature. I write lots of code comments (prefixed with // or # ) and markdown blockquoted paragraphs (prefixed with > ) --- it would be painful if ne couldn't Paragraph them while preserving the left-hand-side prefixins.

Actually, I'd expect that for any series of lines, ne should be able to identify if every line starts with the same prefix (whatever it may be --- different programming languages may use different commenting/quoting/whatever styles), and preserve that pattern when Paragraphing it. For example, I'd expect ne to be able to correctly Paragraph:

// XXX This comment is all kinds of important. I'm not kidding. Please keep reading!
// XXX Make a note of it in your records.
// XXX And
// XXX also watch for detritus.

I'm not sure why ne would be limited to special-cases like // and > .

[utoddl]

Hmm. This is in the category of "historical accident". There was a brief time when the Paragraph command tried to do what you indicated above, and it usually worked out alright. But there were lots of corner cases -- paragraphs that had nothing to do with programming or that contained comment-like characters that weren't comments at all -- which prompted us to revert back to a "dumber but consistent" behavior. Looks like the docs didn't make the transition.

I would like to have a comment-aware paragraphing option, but not at the expense of breaking normal text. Ne is supposed to be a Nice Editor, where unexpectedly mangling your text is decidedly Not Nice. I hope to revisit this issue some day.

[uvtc]

Thanks, Todd. Glad to hear it may be revisited in the future. Super useful feature, IMO, even if it were strictly limited to a subset of prefixes (like "> ", "// ", "# ") or to even just one prefix ("> ") to start with.

Maybe it would be possible for the user to set a default pref that says, "consider only the following strings as formatting prefixes when reflowing text with Paragraph: ...", if that simplifies and tightens things up.

Also, not sure if this is a corner case you were thinking about trying to solve, but I see that in code, lines of code usually directly follow comment lines (with no empty line between them). I could see how making Paragraph smart enough to handle that corner case would be complicated, and I would not expect Paragraph to try and figure that out. Paragraph works on paragraphs (delineated by blank lines). That is, I'd be happy to and expect to change:

// Comment line 1.
// Comment line 2.
function code(goes here) {...}

to

// Comment line 1.
// Comment line 2.

function code(goes here) {...}

before hitting [p on that comment block to Paragraph it, and then go back and remove the extra newline.

[uvtc]

It occurs to me that I could also use the ne Through command, and have a small shell script do this. I'll give that a try.