I'm not sure whether to stay quiet or point out that you may have misread me...
I'm referring to a select choice of words that just happens to neatly fall against the 72-character limit... =) Here's the commit message I was referring to: Like man(7), mdoc(7) is a macro package for marking up computer manuals. The main difference is that mdoc is semantic rather than presentational, providing authors with a full DSL to abstract page markup from low-level Roff commands. By contrast, `man` is minimalist and leaves formatting to the author, who is expected to have a working amount of Roff knowledge. Therefore the use of `mdoc` for marking up Node's manpage is a decidedly better choice than bare `man`. Less room is left for error and mandoc(1) offers very robust error-checking and linting. I've been writing every commit-message like this for years and I got too good at it, now I look completely mental... :-\ On 16 April 2018 at 23:56, Ralph Corderoy <ra...@inputplus.co.uk> wrote: > Hi John, > > > Does anybody else here manage to line-wrap their commit messages at > > *precisely* 72-characters without the aide of hyphenation or > justification? > > ;-) Or is it just me? > > I find myself sometimes breaking a line after a comma or full stop, > without starting a new paragraph, if the line is still a good length. > Rather than vim's `gwap', or fmt(1), say. > > It's convenient for later edits being isolated in their ripple effect. > I think Bell Labs book authors often did this when using ed(1) as the > edits were line based and the edit often wanted to work on a clause. > Does anyone here know a fmt(1)-er that tries to do this? > > fmt(1) alternatives I know are > > https://en.wikipedia.org/wiki/Par_(command) > http://search.cpan.org/~neilb/Text-Autoformat-1.74/lib/Text/ > Autoformat.pm > > The latter can easily be run from the command line, > not just used as a module in Perl. > > -- > Cheers, Ralph. > https://plus.google.com/+RalphCorderoy > >