Quarto website: Set fixed column widths for Arguments section?

[grantmcdermott]

Subtitle: wrapping of long argument lists.

Hi folks,

This is a slightly difficult issue to describe, but easy to visualize. So let me first direct you over to the rendered webpage that prompted my question: https://grantmcdermott.com/tinyplot/man/tinyplot.html#arguments

As you can (hopefully) see, there is a large whitespace gap between the function arguments (x, y, etc.) and their respective descriptions ("the x and y arguments provide the x and y coordinates for the plot.", etc.) Indeed, the description column here is very thin.

If you scroll further down the page, you'll see that the culprit is the very last subset, na.action, drop.unused.levels entry right at the bottom of the Arguments section. This long entry doesn't get wrapped and so all of the explanation text in the next column is pushed over to the right.

Question: Is there are way to enforce max column widths of these respective column, and/or ensure wrapping so that we don't end up with excessive whitespace and squished description columns?

FWIW wrapping automatically takes place in the R console if you look at same help page locally (here: with ?tinyplot in RStudio).

.

Thanks for taking a look and apologies if I'm overlooking something obvious.

[vincentarelbundock]

Quick response before going to work.

1. The nowrap setting is essential for the vast majority of packages, because altdoc used to do the weirdest breaks for single argument names.

2. FWIW, I actually think that tinyplot should split the arguments on separate lines. I know this is Base style, but I personally think it displays badly both online and in the pdf.

3. An immediate solution would be to sed this string out of the man pages before uploading the website: white-space: nowrap;.

For reference, nowrap is added here: https://github.com/vincentarelbundock/altdoc/blob/c50ebef035e81aeb225f68fa27d7e9c86387268a/R/rd2qmd.R#L22

[etiennebacher]

I don't think we have a better solution than point 3. One could override the CSS with white-space: collapse but it would be applied to the whole website (therefore defeating point 1) and I think currently it would be hard to apply it just to man pages, so I'm closing this.

Thanks for the detailed answer, Vincent.

[grantmcdermott]

Thanks both.

Just so I understand, for 2 is the suggestion to duplicate the description for each newline argument? I'm not sure how to make multiple argument entries inherit the same description block otherwise...

I'll think on a custom post processing command to add to the GA workflow file for 3.

[etiennebacher]

For reference, nowrap is added here: vincentarelbundock/altdoc@c50ebef/R/rd2qmd.R#L22

Actually I think we could change nowrap to collapse and it would give this:

@vincentarelbundock do you have some examples in mind where altdoc was weirdly wrapping arg names? I could test those to see if they play nicely with collapse.

[vincentarelbundock]

that looks interesting.

but sorry, no, it was too long ago and I forgot. My guess is packages with dots in variable names will be especially vulnerable

[etiennebacher]

Ok I'll do a mock package to try this out

[grantmcdermott]

Well, tinyplot does have lots of dots in names (a bitter base R convention I had to swallow for the sake of consistency) 😏

So if it doesnt show up there we might be okay.

[etiennebacher]

Here's with #308:

• quarto

• docute

• docsify

• mkdocs

---

With nowrap (current situation), with quarto only:

WDYT?

[etiennebacher]

I'm moving forward with this because it's easier to test on multiple existing packages than keeping the change in a PR. If something breaks, we can always revert it before releasing.

[grantmcdermott]

Super, thanks @etiennebacher. I'll update to the dev version of altdoc for tinyplot and LYK if I notice any red flags.

BTW is the preferred way to do this now (i.e., Suggest a remote package) via R-universe?

[etiennebacher]

BTW is the preferred way to do this now (i.e., Suggest a remote package) via R-universe?

I had no idea one could do that, I always keep a Remotes field only (this also reminds me to check if the dependency has been updated before I do a release).

[grantmcdermott]

Got it, thanks. I just saw that the R-universe approach is failing anyway, since the altdoc build hasn't updated passed 0.5.0.

[grantmcdermott]

(As an aside, the R-universe approach might be a nice solution for tidypolars if you want to get that on CRAN with the polars/neo-polars dependency remaining on R-universe.)

[etiennebacher]

Great idea! That didn't even cross my mind but it would be great to have it on CRAN, thanks 👍

I have to read more about this, it sounds a bit weird to me that a package can be on CRAN and its dependencies on R-universe but not on CRAN. I remember the CRAN team being very sensitive about downloading external dependencies so I'm surprised it's that "easy" to bypass.

[grantmcdermott]

P.S. The updated tinyplot website looks great, thanks! https://grantmcdermott.com/tinyplot/man/tinyplot.html#arguments