commonmark-spec

commonmark-spec copied to clipboard

The text for link reference definitions is a bit hard to comprehend and a bit contradictory. See the discussion here.

Maybe something like this is better?

A link reference definition consists of a link label, optionally preceded by up
to three spaces of indentation, followed by a colon (:), optional spaces or tabs
-(including up to one line ending), a link destination, optional spaces or tabs
-(including up to one line ending), and an optional link title, which if it is
-present must be separated from the link destination by spaces or tabs. No
+(including up to one line ending), a link destination, an optional link title,
+which if it is present must be separated from the link destination by spaces or
+tabs (including up to one line ending), and spaces and tabs. No
further character may occur.

But this one is rather hard to explain in prose as @jgm mentions without a word for [\t ]

But this one is rather hard to explain in prose as @jgm mentions without a word for [\t ]

This is not to take on the commonmark spec which is a very noble and useful endeavour and I respect all the hard work that went into making it happen but:

[rant] I really don't understand why people making standards (whatwg being the ultimate offenders) are so keen on using natural language when we have accessible – in the sense you don't need higher mathematics knowledge – formalisms that make all that precise, concise and reusable across the spec.

The spec is full of repetitions of natural language constructs and hard to distinguish subtleties. For example I had example 621 failing: of course in contrast to here there's no "optional" there. With an ABNF rule for it, the distinction and actual sequencing structure would be pretty obvious. [/rant]

I don't think this is the place for rants but, well, because the BNF for markdown (or html) is:

markdown = .*

(http://trevorjim.com/a-grammar-for-html5/)

@wooorm that's throwing the baby with the water :–) It's not because a BNF for the full format is useless that it is aswell for describing its components.

@dbuenzli I'm sympathetic to your rant. Perhaps it would have been better to do it that way, even though not everything will be describable in that form.