End footnote definition with one blank line.

[raphlinus]

According to the linked issue, a footnote definition should end with a
blank line. This is similar to the rule for lists, which end with two
blank lines. The code previously required two blank lines in both
cases, this patch changes it to just one for footnote definitions.

Fixes issue 20.

[googlebot]

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed, please reply here (e.g. I signed it!) and we'll verify. Thanks.

---

• If you've already signed a CLA, it's possible we don't have your GitHub username or you're using a different email address. Check your existing CLA data and verify that your email is set on your git commits.
• If you signed the CLA as a corporation, please let us know the company's name.

[googlebot]

CLAs look good, thanks!

[raphlinus]

Fixes issue #20, see more discussion there.

[notriddle]

If this is going to be merged, specs/footnote.txt needs to be updated.

[GuillaumeGomez]

I need this to be merged. Do you know when this we'll done?

[raphlinus]

In order to merge this, I need more confidence that it's compliant with the spec, which I currently don't have. Otherwise, if you need non-spec-compliant behavior for compatibility, then divergence from the spec needs to be a flag that gets set.

[GuillaumeGomez]

Seems like it is.

[raphlinus]

Ah, ok. I'll play a bit with whether this PR actually does the right thing, and if so I'll merge it relatively soon.

[GuillaumeGomez]

Great, thanks a lot!

PS: pulldown-cmark is now the markdown parser used in rustdoc (congrats!).

[ScottAbbey]

Seems like it is.

I don't think this shows anything in favor or against. That tool has no support for footnotes so it interprets [^footnote]: Thing as a link to "Thing".

It appears as though the footnote spec in this repository treats them more like link references than list items. This is probably not right because link references can only contain a small number of specific items, while a footnote seems like it should be a container block.

I also just left a comment at #20 (comment).

[raphlinus]

Based on discussion in https://internals.rust-lang.org/t/what-to-do-about-pulldown-and-commonmark/5115 I am inclined to merge this. Any final comments?

Thanks to everybody who's weighed in here, and sorry for not taking action on it sooner.

[GuillaumeGomez]

Finally! \o/

[critiqjo]

Since this would be a "breaking change" anyway, may I suggest breaking a little further and disallowing blocks inside footnote definition (if it is easy enough)? For example:

a [^b].

[^b]: * c
* d

Should produce either:

<p>a <sup class="footnote-reference"><a href="#b">1</a></sup>.</p>
<div class="footnote-definition" id="b"><sup class="footnote-definition-label">1</sup>
* c
* d
</div>

or

<p>a <sup class="footnote-reference"><a href="#b">1</a></sup>.</p>
<div class="footnote-definition" id="b"><sup class="footnote-definition-label">1</sup>
* c
</div>
<ul>
<li>d</li>
</ul>

So that this would just be a baseline implementation of footnotes. And further changes based on whichever spec that we choose to adopt will not "break" much. Since pulldown-cmark is now being used by rustdoc, implementation-specific features should be kept as thin as possible, in my opinion.

[GuillaumeGomez]

Thanks!

[raphlinus]

I'm merging this because it seems like an overall improvement. The suggestion by critiqjo seems valid, but I haven't had time to research it in detail. I'd happily take a patch for it as a further refinement, as long as there's evidence it won't cause regressions, etc.

This is released as v0.0.15.

[ScottAbbey]

This broke several footnotes spec tests:

failures:
    spec_test_3
    spec_test_4
    spec_test_5
    spec_test_7

test result: FAILED. 4 passed; 4 failed; 0 ignored; 0 measured

It's important to pass all these since they are basically the only definition of this particular footnote spec. I haven't looked at them, but I am guessing the tests need to be updated.