Re: 80 char line limit, and we should stop the convention

[Kaz Kylheku]

On 2012-02-27, Xah Lee <xahlee@gmail.com> wrote:
> the 80-char limit in unixes, and the convention to truncate/format

This is one thing you can't pin on Unix.

There is no such limit in "unixes". Some text console windows are that narrow;
others can be resized. 

Unix introduced a bunch of languages, all of them freeform, more or less, and
without the column-oriented junk of punched card languages.

> file no more than 80 chars per line in unix email or source code (such
> as emacs's fill-paragraph), are from punch cards.

I remember reading some quote from one of the Unix "grampas", that fixed
records belong to the age of the 80 column punched card, echoing your very
criticism.  The dbm database API introduced in Unix has arbitrary length
records.

However, it makes sense to limit the number of columns in text files
that are for human consumption like source code or e-mail.

A compiler can read 1000 column text; do you want to, though?

Of course, this limiting to a width does not have to be done by the typist.

> we should stop the habit of formatting lines to 80 chars, in source
> code. e.g. stop using any form of fill-paragraph.

Unless you think that 100,000 and up columns is okay in a source file, you
tacitly acknowledge that there has to be a limit.

> also, am going to stop formatting my elisp function's doc string to 80
> chars.
>
> instead, format them by semantic unit. e.g. by paragraph.

You have good company in the late Erik Naggum, who also once said,
I think, something almost the same. Don't manually break lines.  Write
paragraphs as one long string, and let the software do the breaking
at display time.

That is a fine idea for e-mails, but not so hot for source code.