When a linebreak occurs in a link, the link is not parsed.

[vermiculus]

[really long pieces of text spanning
two lines](link)

should work.

(assert
 (= (md-to-html-string "[long\nline](link)")
    (md-to-html-string "[long  line](link)")))

[yogthos]

Unfortunately, thats a known limitation. The parser currently tokenizes each line individually, so tags spanning multiple lines are not supported.

[vermiculus]

Is this something that's reasonable to be worked-around? It's conceivable that a preprocessor could ensure tags are on one line when appropriate (i.e., when not ending with a double-space) before it's passed to the tokenizer.

[yogthos]

That would be possible, but would also incur a runtime overhead. I would be ok with that as an opt in, the way it's done with the reference style links.

[ir-regular]

I would also be interested in links with linebreaks because of http://www.gravizo.com/ :)

Would requiring a "line continuation" indicator (like a backslash in bash) make it easier to handle? You could check the last character, read in all lines that have a continuation indicator + 1 more line, concatenate and put through normal link handling.

[yogthos]

Yeah, you'd have to implement something like that. The tricky part is handling that intelligently within lists, especially when you use paragraphs for list items, and nested lists.