Further Minors.

[SebKrantz]

Not by any means wanting to keep anyone busy here. I just noticed that some recent commits introduced a weird rendering of the usage section of collapse::pivot() (transpose argument): Docs | Source.

I believe as late as 3af98ac this was not the case.

Then, I finally wondered why an empty line is placed above code chunks provided in NEWS.md (Source).

Many thanks for any further consideration!

[hadley]

I feel like I must be missing something because both of these look fine to me:

Can you be a bit more explicit about the problems you're seeing?

[SebKrantz]

Thanks. So for pivot, the site looks like this

Whereas the docs looks like this

Of course I understand that such commentary of arguments is non-standard, but before the changes regarding the replacement functions pkgdown also rendered the pivot docs properly.

Regarding the NEWS, there is an empty line before the code begins. In all such chunks in the NEWS. Not only in collapse, also in fixest for example.

[hadley]

Oooh, got it. I'll take a look. For future reference, since these are unlikely to have the same underlying cause, it would be useful if you created two separate issues.

[hadley]

The code blocks in https://pkgdown.r-lib.org/dev/news/index.html all look ok, but they are non-R code. ... The space looks like it's the same size as the copy button but I think that's a red herring ... It looks like it's not coming from css, but because there's an extra empty line. So presumably something going wrong in the html highlight tweaking, but I didn't think I'd touched that code recently.

[hadley]

The messed up usage comes from #2626; I'd forgotten about the case where the final line in a one-line-per-argument usage will be parseable with a non-syntactic call to =. Maybe I can just special case =, but there might be other cases with multi-line default args where this also comes up.

[hadley]

Yeah, the news problem is because we're generating:

<div class="sourceCode" id="cb1"><pre class="downlit sourceCode r">
<code class="sourceCode R"><span></span>
<span><span class="st">"x"</span></span></code></pre></div>
</div>

instead of

<div class="sourceCode" id="cb1"><pre class="downlit sourceCode r">
<code class="sourceCode R"><span><span class="st">"x"</span></span></code></pre></div>
</div>

It's something to do specifically with syntax highlighting for R because you also don't see the gap if the R code doesn't parse.

[hadley]

The problem goes away if I comment out downlit::downlit_html_node(html), but the syntax highlighting stays. That suggests downlit is getting double called. But why?

[hadley]

Minimal reprex:

raw <- '<pre class="downlit sourceCode r">
<code class="sourceCode R"><span><span class="st">"x"</span></span></code></pre></div>'

html <- xml2::read_html(raw)

downlit::downlit_html_node(html)
cat(as.character(html))

....

I see the same problem in pkgdown 2.0.9, so doesn't look like a new issue.

[hadley]

Same problem with pkgdown 2.0.0 and downlit 0.4.0 so this has definitely been around for a while.

[hadley]

Phew. That was a journey of discovery.

[SebKrantz]

Thanks @hadley, much appreciated! collapse's site looks great now. Cheers, Seb.