Remark: Link parsing is more lenient than either GitHub or commonmark.

[samouri]

summary
remark-parse is currently pretty lenient about whitespace within links, more so than either GitHub or commonmark. Here is an example of text that I would not expect to become a link, yet currently does:

[display text] (https://example.com)

In remark-parse, the (https://example.com) text would be considered the address of the link, even though there is a space between the two parts.

Is this intended? It feels like a bug. In particular this has been frustrating to me in its integrations with prettier, where I often am writing markdown intended for GitHub and the whitespace between the two parts is automatically stripped.

If y'all consider this a bug, I'd be happy to submit a PR to fix it!

[ChristianMurphy]

@samouri are you seeing this even when the option commonmark is set to true?

[samouri]

commonmark doesn't seem to have an effect on this. The pedantic options does net the behavior I'd expect (separating the text and link)

[wooorm]

Reason for that is that GH used to support the space, but “pedantic” didn’t.

If you’re willing to work on the problem, we would accept PRs with a fix.
It’s also something that would go in https://github.com/remarkjs/remark/issues/439!

[samouri]

What is the usual protocol for informing downstream folks of this change?

[wooorm]

@samouri How do you mean? What are you worrying/wondering about?

[samouri]

I know that prettier is dependent on this package. This will mean a different behavior when applying prettier to markdown files with this edgecase.

I think it's a good change, but I figure any change at all could be disruptive for dependents (e.g. if they have some unit tests that start failing).

[wooorm]

This is arguable* a breaking change, yes. Whether a release is breaking is signalled through semver.

* arguable because, if someone before wrote [a] (b), and prettier formatted it as [a](b), and we now stop supporting the earlier. Everything’s pretty fine 🤷‍♂️

But indeed, remark-parse is used by millions of users, and we don’t want to break their flow. We’re generally on the safe side with releases.

[samouri]

Thank you for the explanation! Makes sense 👍